Biodiversity conservation in traditional farming landscapesopus.uni-lueneburg.de/opus/volltexte/2015/14348/pdf/Dorresteijn... · identifying system properties that facilitate biodiversity

Biodiversity conservation in

traditional farming landscapes

The future of birds and large carnivores in Transylvania

Doctoral thesis by Ine Dorresteijn

Biodiversity conservation in

traditional farming landscapes


Academic dissertation Submitted to the Faculty of Sustainability of Leuphana University

for the award of the degree

‘Doctor of Natural Sciences’

-Dr. rer. nat.-

Submitted by

Ine Dorresteijn

Born on 15.05.1984 in Utrecht (The Netherlands)

Date of submission: 09.01.2015

Doctoral advisor and reviewer: Prof. Dr. Joern Fischer

Reviewer: Prof. Dr. Tobias Kuemmerle

Reviewer: Prof. Dr. Kristoffer Hylander

Date of disputation: 29 May 2015

Copyright notice

Chapters II-IX, and Appendix I & II have either been published or are in preparation for

publication in international scientific journals. Copyright of the text and the illustrations is with the

author or the authors of the respective chapter. The publishers own the exclusive right to publish

or to use the text and illustrations for their purposes. Reprint of any part of this dissertation

requires permission of the copyright holders.

Photo credits: Jan Hanspach (cover page; page 14, Fig 1.2a,b,d), Nathanaël Vetter (page 33), and

Rémi Bigonneau & Anne-Catherine Klein (page 155, 187), all other photos by Ine Dorresteijn

© Ine Dorresteijn (Chapters I, II, V & IX)

© Springer (Chapter III & IV)

© Elsevier (Chapters VI & Appendix II)

© Public Library of Science (Chapters VII & Appendix I)

© International Association for Bear Research and Management (Chapter VIII)

Author΄s address:

Leuphana University, Faculty of Sustainability Science

Scharnhorststrasse 1, 21335 Lueneburg, Germany

e-mail: [email protected]

mailto:[email protected]

Acknowledgements

‘Even the woodpecker owes his success to the fact that he uses his head and keeps pecking away

until he finishes the job he starts’ – Coleman Cox

I quite often felt like this woodpecker during my PhD, and I would still be pecking away if it wasn’t

for the many people that helped from the start to the finish of my PhD.

First of all I wish to thank my supervisor Joern Fischer. This PhD would not have been

possible without him. Joern supported, encouraged, and inspired me, and he has given me

irreplaceable guidance and advice at every stage of the PhD. Thanks for sometimes keeping me on

the ground, but also for giving me the space to carry out my ideas, even if that meant more time in

the field.

I am grateful to the entire team for their support during my PhD, but most of all just for

making me part of the team. Jan Hanspach provided important guidance throughout the PhD.

Many thanks for helping, listening, and handling my impatience, especially with R. Also thanks to

Henrik von Wehrden for helping me through several statistical and GIS problems. I learned a lot

about the region from Tibor Hartel, but also had a great time with him during my first month in

the field. Thanks to Dave Abson for proofreading the first chapter of this dissertation, and

together with Julia Leventon and Heike Zimmermann for creating a nice atmosphere and many

serious and not so serious discussions. Thanks to my fellow PhD-students Jacqueline Loos, Andra

Milcu, Friederieke Mikulcak, and Marlene Roellig for sharing all the ups and downs, frustrations

and laughs, invaluable discussions, chocolates and cookies, and for being my friends during the past

years. I also had the pleasure to work with two bright bachelor students. Being able to discover the

world of identifying birds by their song together with Lunja Ernst was incredibly helpful. Thanks to

Lucas Teixeira for helping me with data entry, data analyses, and making fieldwork at night look

easy and fun – it was great having you around for so long.

This dissertation arises from a large amount of data collected in the field. None of this

would have been possible without the help of the most awesome field team, and a big thanks to all

of you! Cathy Klein spent nine months with me in the field and deserves special thanks. Having

such a reliable, hardworking, and funny good friend by my side was the best recipe for two

successful field seasons. Joris Tinnemans was tireless in the field and while sorting camera pictures

and with his humour made both years a lot of fun. Rémi Bigonneau did a fantastic job on the bird

surveys and made the best barbecues for the team. Arpad Szapanyos was indispensable at several

stages during my time in Romania and I could always count on him. Thanks to Cosmin Moga for

helping with the bird surveys when the migrants arrived, but also for the wines in the evenings.

Hans Hedrich did an incredible job with the interviews and translations and had the superb skill of

making people feel comfortable during the interviews. Annamarie Krieg, you were a sunshine in

the field house even after long days of anthill counting in the heat. Ben Scheele and Alexander

Fletcher were crazy enough to spend their holidays counting anthills and I very much appreciate

http://www.searchquotes.com/quotation/Even_the_woodpecker_owes_his_success_to_the_fact_that_he_uses_his_head_and_keeps_pecking_away_until_/12759/

http://www.searchquotes.com/quotation/Even_the_woodpecker_owes_his_success_to_the_fact_that_he_uses_his_head_and_keeps_pecking_away_until_/12759/

their efforts. Also thanks to Jenny Long who spent part of her holidays setting out camera traps

and sorting pictures. Martin Bubner and Frank Dietrich were a hardworking and dependable team,

and fortunately also knew how to fix the lights in our field house. Nate Vetter and Frans Meeuwsen

reinforced the team with new energy in the last month of fieldwork and without their help we

would not have been able to finish our ambitious camera trap plan. Thanks to Kuno and

Annemarie Martini for opening their house to us in Sighisoara.

I am especially grateful to all respondents that participated in the interviews, the people

whose land we accessed for our fieldwork, and the people that helped with directions, cars, or any

other problem encountered in the field.

I am also thankful to our partner organizations in Romania. I enjoyed the collaboration

with The Milvus-group in Targu Mures very much, and would like to thank all the people that

collected data on bear activity. In particular, I would like to thank Attila Kecskés, Hana Latková,

Zsófia Mezey, and Szilárd Sugár, with whom I worked closely for Chapter IV of this dissertation.

My thanks also go to the Mihai Eminescu Trust, especially Caroline Fernolend and Oliviu Marian,

for their help and logistical support.

I would like to thank all my co-authors for their help with the preparations of the different

manuscripts. I would especially like to thank Euan Ritchie and Dale Nimmo. Euan made it possible

for me to spend a month in his working group at Deakin University to learn more about trophic

ecology and have an overall amazing experience in Australia. Thanks to Dale for his valuable input

on several manuscripts and also for letting me live in his house while I was in Australia. Thanks to

Laura Kehoe and Tobias Kuemmerle for providing extra camera traps and driving them all the way

down to Romania.

This research would not have been possible without funding. I am grateful to the

Alexander von Humboldt Foundation and the International Association for Bear Research and

Management for funding our research. Most of the research was funded through a Sofja

Kovalevskaja Award by the Alexander von Humboldt Foundation to Joern Fischer. The

International Association for Bear Research and Management funded the interview work and made

this interdisciplinary dissertation possible.

Finally, thanks to all friends and family that reminded me of also having a life next to the

PhD. Thanks to the Schultner family for their interest and support, and for providing me with the

most delicious energy packages when times were busy. A warm thanks go to my brother, sister, and

sister-in-law – Bram, Marleen, and Dorien Dorresteijn – for always showing interest in my work

and their visits to Germany. I am grateful to my parents, Leo and Joyce Dorresteijn for their

support and encouragements, but also for visiting me wherever I decide to go, and for moving my

stuff to those places. A loving thanks to Jannik Schultner who was my rock during my PhD years,

kept me sane at especially more stressful times, let me rant when I wanted too, but also made this

time wonderful!

Preface

This dissertation is presented as a series of manuscripts based on empirical research carried out in

Transylvania, Romania. Chapter I provides a general overview of the dissertation, including the

overarching goal and specific aims, a summary of all included manuscripts, a synthesis of the results

identifying system properties that facilitate biodiversity conservation in traditional farming

landscapes, and finally an outlook for conservation priorities in traditional farming landscapes.

Beyond Chapter I, the manuscripts included in this dissertation (Chapters II-IX) are divided into

three sections (A, B, and C). Section A examines the effects of local and landscape scale land-use

patterns on birds and large carnivores and how these animals may be affected by future land-use

change (Chapters II-V). To gauge the role of traditional land-use elements for biodiversity, in

Section B, I focus on wood pastures as one prominent example of such traditional elements

(Chapters VI-VIII). Lastly, in section C, I use a social-ecological systems approach to understand

social drivers underlying human-bear coexistence (Chapters IV and IX). With the exception of

Chapter II, all manuscripts are either published, in revision, or under review in international

scientific journals. I, the author of this dissertation, conducted the majority of the research

presented in this dissertation and am the lead author of the manuscripts presented in Chapters II-

V, VII, and IX. I provided important contributions to Chapters VI and VIII as a co-author. A

reference to the journal each manuscript is submitted to and the contributing co-authors is

presented on the title page of each chapter. The Appendix contains two published manuscripts that

I co-authored during my PhD. The research of these two manuscripts was conducted in the same

study area but with a focus on butterflies and plants; their similar research context provides

additional insights for the general synthesis of this dissertation.

Table of contents

Abstract 1

Chapter I 3

Biodiversity conservation in traditional farming landscapes:


Section A: Land-use effects on biodiversity 29

Chapter II 31

Maintaining bird diversity in a traditional farming

landscape in Eastern Europe

Chapter III 53

Impact of land cover homogenization on the corncrake

(Crex crex) in traditional farmland

Chapter IV 71

Human-carnivore coexistence in a traditional rural

landscape

Chapter V 105

Incorporating anthropogenic effects into trophic ecology:

Predator-prey interactions in a human-dominated landscape

Section B: Traditional wood pastures 129

Chapter VI 131

Wood-pastures in a traditional rural region of Eastern

Europe: Characteristics, management, and status

Chapter VII 153

The conservation value of traditional rural landscapes:

The case of woodpeckers in Transylvania, Romania

Chapter VIII 169

Brown bear activity in traditional wood-pastures in Southern

Transylvania, Romania

Section C: Human-bear coexistence 183

Chapter IX 185

Social factors mediating human-carnivore coexistence:

Understanding coexistence pathways in Central Romania

(Chapter IV is also part of this Section)

References 213

Appendix I 239

Low-intensity agricultural landscapes in Transylvania

support high butterfly diversity: Implications for conservation

Appendix II 265

Plant diversity in a changing agricultural landscape mosaic

in Southern Transylvania (Romania)

1

Abstract

Traditional farming landscapes typically support exceptional biodiversity. They evolved as tightly

coupled social-ecological systems, in which traditional human land-use shaped highly

heterogeneous landscapes. However, these landscapes are under severe threats of land-use

change, in particular through land-use intensification and land abandonment. Changing land-use

practices fundamentally change the structure of traditional farming landscapes, which implies a

direct threat to the biodiversity they support. Navigating biodiversity conservation in such changing

landscapes requires a thorough understanding of the drivers that maintain the social-ecological

system.

This dissertation aimed to identify system properties that facilitate biodiversity

conservation in traditional farming landscape, focusing specifically on birds and large carnivores in

the rapidly changing traditional farmland region of Southern Transylvania, Romania. In order to

identify these properties, I first examined the effects of local and landscape scale land-use

patterns on birds and large carnivores and how they may be affected by future land-use change.

Bird diversity was supported by the broad gradients of woody vegetation cover and

compositional heterogeneity. Land-use intensification, and hence the loss of woody vegetation

cover and homogenization of land covers, would thus negatively affect biodiversity. This was

especially evident from predictions on the distribution of the corncrake (Crex crex) in response

to potential future land cover homogenization. Here, a moderate reduction of land cover

diversity could drastically reduce the extent of corncrake habitat. Further results showed that

the brown bear (Ursus arctos) would mainly be affected by land-use change through the

fragmentation of large forest blocks, especially if land-use change would reduce habitat

connectivity to the presumed source population in the Carpathian Mountains. Moreover, this

dissertation revealed that large carnivores (brown bear and wolf, Canis lupus) may have

important and often ignored roles in structuring the ecosystem of traditional farming

landscapes by limiting herbivores.

Second, to gauge the role of particular traditional land-use elements for biodiversity

this dissertation focused on the conservation value of traditional wood pastures, as one

prominent example of such traditional elements. Wood pastures were found to have a high

conservation value. The combination of low-intensity used grasslands with old scattered trees

provided important supplementary habitat for different forest species such as woodpeckers

and the brown bear. Worryingly, current management of wood pastures differed from traditional

techniques in several aspects, which may threaten their persistence in the landscape.

Third, this dissertation took a social-ecological systems approach to understand how links

between the social and ecological parts of the system affect human-bear coexistence. The majority

of people had a positive perception on human-bear coexistence. The use of traditional sheep

2

herding techniques combined with the tolerance of some shepherds to occasional livestock

predation facilitated coexistence in a region where both carnivores and livestock are present. More

generally, the genuine links between people and their environment (i.e. where people value their

natural surroundings) were important drivers of people’s positive views on human-bear

coexistence. However, perceived failures of top-down managing institutions could potentially erode

these links and reduce people’s tolerance towards bears.

Through the consideration of two different animal taxa, this dissertation revealed six

important system properties facilitating biodiversity conservation in traditional farming landscapes.

Biodiversity was supported by the heterogeneous character of the traditional farming landscape at

multiple spatial scales. At the scale of the study area, similar proportions of the main land-use types (arable

land, grassland, and forests) support species associated with farmland as well as with forests,

through for example habitat connectivity and continuous spill-over between land-use types.

Heterogeneous landscapes can further support biodiversity through complementation and

supplementation of habitat at the landscape scale, where species can occur outside their considered core

habitat. Gradients of woody vegetation cover and heterogeneity, supported biodiversity at both local and

landscape scales, possibly through the provision of a wide range of resources or by facilitating

cross-habitat movements and spill-over. The heterogeneous character of the landscape is tightly

linked to traditional land-use practices, and thus these practices are key to maintaining biodiversity. In

addition, specific traditional land-use elements such as wood pastures have high conservation value,

while specific practices such is traditional livestock husbandry techniques facilitate human-

carnivore coexistence. Top-down limitation of large carnivores on herbivores may facilitate biodiversity

conservation through for example enhancing vegetation growth and tree regeneration. The genuine

links between humans and nature supported human-bear coexistence, and these links may form the core

of people’s values and sustainable use of natural resources.

Maintaining or preserving these six system properties should be a priority for

biodiversity conservation in traditional farming landscapes. However, to accomplish this, there

is an urgent need to develop more holistic visions for biodiversity conservation in traditional

farming landscapes that integrates the entire social-ecological system. Such a holistic approach may

comprise ‘broad and shallow’ landscape-scale conservation measures targeting the heterogeneous

landscape character of the forest-farmland mosaic at multiple spatial scales. Large scale

conservation measures may be complemented with more ‘deep and narrow’ conservation measures

targeting specific species, land-use types, threats, or traditional practices. Finally, conservation

measures should encourage the integration of the entire social-ecological system by recognizing and

incorporating important links between people and the environment. Traditional farming landscapes

are rapidly disappearing worldwide and developing conservation visions to navigate such landscape

through land-use change are now needed to prevent major biodiversity declines in these landscapes.

3

Chapter I

4

5

Chapter I

Biodiversity conservation in traditional farming landscapes


‘The most effective way to save the threatened and decimated natural world is to cause people to fall in love

with it again, with its beauty and its reality.’ – Peter Scott

http://www.goodreads.com/author/show/314707.Peter_Scott

Chapter I

6


7

Introduction

Traditional farming landscapes often harbour exceptional biodiversity and have high conservation

value. These landscapes evolved as tightly coupled social-ecological systems, which are now under

severe pressure of land-use change. From a biodiversity conservation perspective this is worrying

because land-use change may potentially erode the high levels of diversity these landscapes support.

Navigating biodiversity conservation in such changing landscapes requires a thorough

understanding of the drivers that maintain the social-ecological system. In this dissertation I focus

on the ecological part of the system and aim to identify system properties that facilitate biodiversity

conservation in traditional farming landscapes, focusing specifically on the traditional farmland

region in Southern Transylvania.

Human impacts on ecosystems

Humans have shaped and impacted the natural environment for tens of thousands of years (Smith

2007). Humanity’s dominance over virtually all ecological systems has created the urgent need to

understand the impacts and consequences of this anthropogenic influence (Vitousek et al. 1997;

Ellis & Ramankutty 2008). Human influences on the environment range from hunting-gathering

activities to the modification of entire ecosystems, most notably through agriculture. Until the

onset of industrialization, relatively low human population densities and limited technological

progress constrained human development. However, with the dramatic population increase and

technological advances of the last 200 years, human influences on the environment are now so

pervasive that they have become the major drivers of environmental change, taking us into a new

geological era, the Anthropocene (Crutzen 2002; Steffen et al. 2007). The Anthropocene is

characterized by a series of rapid biophysical and socio-economic changes (i.e. global change) that

are threatening ecosystems and human well-being. For example, during the last 50 years, the

structure of many ecosystems changed more rapidly than at any time in history (Millennium

Ecosystem Assessment 2005; Steffen et al. 2007). One of the most notable consequence is the

global biodiversity decline, with current rates potentially leading to the sixth mass extinction event

(Pimm et al. 1995; Pereira et al. 2010; Barnosky et al. 2011; Monastersky 2014). This rapid loss of

biodiversity matters not just because of the intrinsic values ascribed to biodiversity, but also

because current rates of biodiversity loss cannot be sustained without substantially eroding

resilience of ecosystems around the world – that is their ability to continue functioning in the face

of external shocks (Folke et al. 2004; Rockström et al. 2009). Ambitious goals to reduce human-

induced biodiversity loss (e.g. the Convention on Biological Diversity) have not yet been reached

(Butchart et al. 2010), and because of this failure, such goals have sometimes simply been

postponed into the future (www.cbd.int). Thus, a central challenge to humanity is to understand,

address and act on the underlying causes of human-induced biodiversity loss to safeguard

biodiversity in the future.

Chapter I

8

Land-use change and biodiversity loss

In terrestrial ecosystems, land-use change has been one of the major drivers of biodiversity loss -

although other drivers such as climate change, pollution, invasive species, and the synergistic

effects between different drivers also significantly contribute to global biodiversity loss (Sala et al.

2000). Land-use change can transform land through the conversion of the natural environment into

farmland or farmland can be transformed through a change in agricultural practices (Foley et al.

2005). Responses of biodiversity to land-use change are complex and dynamic and depend on the

type of land-use change and the ecological setting (DeFries et al. 2004). The conversion of the

natural environment into farmland caused biodiversity loss worldwide (Foley et al. 2005). Cropland

and pasture are now the largest terrestrial biome occupying 40 % of the land surface (Foley et al.

2005), while the area of forests has been halved over the past three centuries (Millennium

Ecosystem Assessment 2005).

Two of the most distinct effects related to the conversion of the natural environment into

agricultural land are habitat loss and habitat fragmentation for many species (reviewed in Fahrig

2003; Fischer & Lindenmayer 2007), especially for birds and mammals (Andrén 1994; Monastersky

2014). The expansion of agricultural land induced a global loss of birds to between a fifth and a

quarter relative to estimates of pre-agricultural numbers (Gaston et al. 2003). For large carnivores,

typical life-history traits such as large body size, large area requirements, slow reproductive rates,

and low population densities make them exceptionally vulnerable to habitat loss and fragmentation

(Crooks 2002). Increased human-carnivore conflicts and a resulting persecution by humans pose

additional threats to large carnivores in modified landscapes (Treves & Karanth 2003). Habitat loss

and fragmentation in combination with human persecution have severely reduced large carnivore

populations worldwide, which on average occupy only 47% of their historical distribution range

(Breitenmoser 1998; Woodroffe 2000; Ripple et al. 2014). However, large carnivores play critical

roles in structuring ecosystems through inducing trophic cascades, and the loss of large carnivores

has caused undesirable changes to a diverse range of ecosystems and its associated biodiversity

(Estes et al. 2011; Ripple et al. 2014).

Farmland biodiversity, in contrast, is to a large degree adapted to and dependent on the

continuation of agricultural management (Tscharntke et al. 2005). Farmland is the dominant

ecosystem in many places (e.g. in Europe) and holds a large part of the world’s biodiversity

(Pimentel et al. 1992). Yet, also farmland biodiversity can be adversely affected by land-use change,

most notably land-use intensification and land abandonment (Tscharntke et al. 2005; Queiroz et al.

2014; Uchida & Ushimaru 2014). Hence, understanding how land-use change affects biodiversity in

agricultural landscapes is important to mitigate global biodiversity loss.

Land-use change in European farmland

In Europe, most forests were cleared for agricultural land before the onset of the Anthropocene

(Kaplan et al. 2009). This historic forest loss, in combination with direct persecution, caused


9

dramatic declines of large carnivore populations in large parts of Western Europe during the 18th

and 19th century (Breitenmoser 1998). Biodiversity loss as a consequence of agricultural

intensification and land abandonment, in contrast, occurred mainly over the last decades and are

now the major threats to European farmland biodiversity (Donald et al. 2001; Tilman et al. 2001;

Stoate et al. 2009).

Agricultural intensification, with the aim to increase agricultural production, has caused

declines of farmland biodiversity and resulted in a halving of the European farmland bird

population (Donald et al. 2001; Voříšek et al. 2010). This decline is caused by ecosystem changes

during intensification due to increased mechanization, increased use of agrochemicals (i.e. fertilizers

and pesticides), loss of crop varieties and management techniques such as crop rotation and

intercropping, and the decline of low intensity land-use. Overall, intensification usually entails

homogenization (e.g. decrease in land cover diversity and woody vegetation cover) at multiple

spatial scales, including entire landscapes (Benton et al. 2003; Tscharntke et al. 2005). These

changes lead to decreased food availability and increased habitat loss and fragmentation for a wide

range of species (Hinsley 2000; Weibull et al. 2003; Concepción et al. 2008; Guerrero et al. 2012).

Species persistence in agricultural landscapes is highly dependent on landscape structure (Fischer &

Lindenmayer 2007), and maintaining or restoring heterogeneity in agricultural landscapes has been

suggested as one of the major strategies to halt farmland biodiversity declines (Benton et al. 2003).

For instance, negative effects of habitat loss and fragmentation may be reduced if the landscape

contains elements that provide habitat connectivity for a range of species, including birds and

carnivores (Uezu et al. 2005; Donald & Evans 2006; Crooks et al. 2011).

In addition to intensification, abandonment of farmland is also becoming more pervasive

globally, especially in marginal agricultural areas characterized by low-intensity farming and

generating relatively low yields (MacDonald et al. 2000; Queiroz et al. 2014). Land abandonment

typically changes the landscape by transforming agricultural land into shrubland, which eventually

turns into forest (Rudel et al. 2005). Land abandonment is often viewed as a negative process for

biodiversity because farmland biodiversity in low-intensity farming regions is often higher than in

natural forests (Höchtl et al. 2005; Lindborg et al. 2008). Indeed, especially in Europe, land

abandonment has been reported to negatively affect a range of taxa, with especially detrimental

effects on farmland birds (reviewed by Queiroz et al. 2014). On the other hand, land abandonment

has also been viewed to offer unique opportunities to restore the biodiversity of natural forest

ecosystems including large carnivores (Navarro & Pereira 2012).

Although intensification and more recently also abandonment have been exacerbated by

the European Union’s Common Agricultural Policy (CAP), the recognition of the value of

farmland biodiversity sparked national and international conservation measures to halt biodiversity

loss on farmland (Young et al. 2005; Henle et al. 2008). Despite these measures, farmland

biodiversity continues to decline and their effectiveness is questioned (Kleijn et al. 2011; Pe'er et al.

2014). Ongoing biodiversity decline highlights the need for more effective biodiversity

Chapter I

10

conservation in Europe’s agricultural landscapes. In this context, traditional farming landscapes are

of great interest to conservation because they often harbour exceptional biodiversity (Tscharntke et

al. 2005; Kleijn et al. 2009).

Traditional farming landscapes: values and challenges

Traditional farming landscapes are increasingly valued for their natural and cultural heritage. Their

importance for biodiversity has been noted worldwide (Ranganathan et al. 2008; Takeuchi 2010;

Robson & Berkes 2011; Liu et al. 2013), including in the traditional village systems of Eastern

Europe (Palang et al. 2006; Fischer et al. 2012). Traditional farming landscapes are characterized by

a long history of relatively persistent farming practices. Farming techniques in these landscapes are

often of low-intensity with low levels of agro-chemical input and little mechanization, that is, a high

degree of manual labour (Bignal & McCracken 2000; Plieninger et al. 2006). This way of farming

has created mixed farming landscapes with a mosaic of different land-uses including specific

traditional landscape elements like wood pastures (Plieninger & Schaar 2008), high land cover and

structural heterogeneity, and relatively abundant semi-natural vegetation (Plieninger et al. 2006).

Another distinctive feature of traditional farming landscapes is that they are often tightly

coupled social-ecological systems, that is, systems in which rural communities influence the

ecosystems and vice versa (Folke 2006). Moreover, the long history of interactions within this

system has created the opportunity for the different entities of the system to constantly co-evolve

(Liu et al. 2007). People have shaped the ecosystem through their activities, such as land-use, and

the ecosystem in turn provided people a variety of ecosystem services (i.e. the benefits people

derive from nature; Millennium Ecosystem Assessment 2005). These ecosystem services range

from provisioning services such as crops, water, and firewood, to cultural services such as the

feeling of a ‘sense of place’, and have historically provided direct incentives for sustainable land-use

(Fischer et al. 2012; Hartel et al. 2014). It is these centuries of co-evolving interactions between

humans and the natural environment that created the high cultural and natural value of the

landscape (Bignal & McCracken 2000). Moreover, landscapes shaped by traditional farming

practices harbour many of the habitats and species that are valued for biodiversity today (Halada et

al. 2011).

Despite their unique natural and cultural values, traditional farming landscapes are under

increasing pressure from modernization and globalization. Rapid socio-economic, political, and

cultural changes lead to the cessation of traditional farming practices in exchange for more

intensive practices, or abandonment of farmland altogether (Henle et al. 2008). The persistence of

these landscapes depends on how the social-ecological system navigates these profound changes

while simultaneously fostering biodiversity conservation and human well-being. Nevertheless,

current policies often fail to acknowledge the links between the social and the ecological parts of

the system, and policies usually target either the social or the ecological part exclusively (Fischer et

al. 2012). Such one-sided policies potentially erode the established historical connections between


11

people and the land that maintain the character of these landscapes, and hence the structures

supporting biodiversity. In addition, in Europe in particular, existing policies for farmland

biodiversity are often poorly adapted to traditional farmland (Sutcliffe et al. 2014). This can be

partly ascribed to a significant research gap, with the majority of European studies on farmland

biodiversity conducted in (Western European) countries with more intensively used farmland,

while biodiversity patterns in the more low-intensity traditional farmland regions (e.g. in Eastern

Europe) remain poorly understood (Baldi & Batary 2011; Tryjanowski et al. 2011).

Therefore, fostering biodiversity conservation of traditional farming regions requires a

thorough understanding of the drivers that maintain the social-ecological system. While this

dissertation does not aim to understand the entire social-ecological system, it deals with some

important aspects of the ecological system while acknowledging the significance of the other

features. In particular, I aimed to identify system properties that facilitate biodiversity conservation

in traditional farming landscapes, focusing specifically on birds and large carnivores in the rapidly

changing traditional farmland region of Southern Transylvania, Romania.

Transylvania’s traditional farmland region

Southern Transylvania in Central Romania (Fig. 1.1) is one of Europe’s last regions that is

dominated by traditional, small-scale farming systems. The study region was shaped by the culture

and land-use of the Saxons, which settled in Transylvania in the 12th and 13th century. The Saxons

came from different German-speaking European countries, and were the dominating ethnic group

in the study region where Hungarians, Romanians, and Roma were also present. Saxon land tenure

was based on communal management of pastures and forests, with individually owned arable fields

(Sutcliffe et al. 2013). The rise and fall of communism influenced land-use in the region. During

communism, agricultural land became collectivized under state ownership, but intensification was

not severe enough to fundamentally change the landscape and its associated biodiversity. After the

fall of communism in 1989 there was a large exodus of Saxons from the region as many resettled to

Germany. This exodus led to an abandonment of part of the agricultural land, while restitution of

small parcels of arable land to the remaining and new population prevented intensification of

agriculture and stimulated semi-subsistence farming.

Despite these social changes, the characteristics of Southern Transylvania’s farming

landscape have changed relatively little since pre-industrial times, and traditional semi-subsistence

farming has maintained a land cover mosaic of relatively similar proportions of forest (28%), arable

land (37%), and grassland (24%; Fig. 1.1). Land-use is primarily determined by topography with

forests occupying the hill-tops, arable fields being located mainly in the valleys, and pastures

occurring on the slopes. Forests are dominated by hornbeam (Carpinus betulus), oak (Quercus sp.),

and beech (Fagus sylvatica). Arable lands are characterized by farming techniques that are small-scale

(most fields are smaller than two hectares) and are of low-intensity (most fields have low chemical

input and are tilled manually). The semi-natural pastures are grazed by sheep (dominant livestock),

Chapter I

12

goats and cattle. Hay meadows provide fodder for livestock and are often harvested by hand. The

lack of large-scale mechanization and lack of homogenization has facilitated a high structural

diversity, as evident for example in different sward heights during spring and summer and a high

abundance of hedgerows, streamside vegetation and scattered trees. Overall, these features

maintain a uniquely high level of biodiversity across multiple taxa in the landscape (Cremene et al.

2005; Akeroyd & Page 2007; Wilson et al. 2012).

Furthermore, the region is still rich in traditional land-use types like the grazing systems of

wood pastures. Wood pastures were established through ancient silvo-pastoral practices, often

consist of open grasslands with scattered old trees, and are one the oldest European land-use types.

This combination of the high ecological value of scattered trees with structural attributes

resembling old-growth forests may support biodiversity of both open-country and forest species

(Bauhus et al. 2009; Fischer et al. 2010). Apart from forest, arable land and grassland, the landscape

mainly harbours scattered small villages (i.e. with usually less than 1000 inhabitants) and orchards.

Part of Southern Transylvania’s farmland is protected through Europe’s largest lowland protected

area, the Târnava Mare - Podisul Hârtibaciului area, within the Natura 2000 network, although no

management plan is in effect to date. However, profound on-going societal and economic changes

since Romania’s inclusion into the EU in 2007 are leading to significant land-use changes in

Southern Transylvania (Mikulcak et al. 2013; Fig. 1.2). Because traditional farming techniques have

become economically unviable, the rural population increasingly either intensifies or abandons

land. These changes, in turn, may significantly impact biodiversity in the future.

Figure 1.1 The study area was located in the foothills (230 m to 1100 m above sea level) of the Carpathian

Mountains in Southern Transylvania, Romania (right). The picture shows the main land-use types of the

study area (left).

Birds and large carnivores in Southern Transylvania

Different animal taxa may be affected by land-use change in distinct ways, and thus working on

multiple taxa may provide a more complete picture of land-use change effects. The traditional

farmlands of Transylvania harbour high densities of different farmland bird species, including

several protected and rare species such as the corncrake (Crex crex) and several woodpecker species

(I. Moga, unpbl. data; Moga et al. 2010). The current trends in land-use intensification and land

abandonment in this region, however, may cause farmland bird declines comparable to those


13

observed in the more intensified countries of Western Europe (Voříšek et al. 2010). Considering

the effects of land-use change on the entire bird community beyond protected species is highly

necessary since common European birds are declining at high rates while populations of rare

species are slowly increasing (Inger et al. 2015).

Unlike most European countries, Romania sustains large and stable populations of large

carnivores (Salvatori et al. 2002), and the study area harbours relatively high densities of the brown

bear (Ursus arctos) and lower densities of the wolf (Canis lupus). In contrast to birds, large

carnivores may be affected especially by changes in forest cover, and may thus be positively

affected by land abandonment. In addition, large carnivores may not only be affected by land-use

or land-use change, but also by the tolerance levels of the rural population towards carnivores. The

reliance of people on forest products (e.g. firewood; Hartel et al. 2014), and the use of traditional

practices such as shepherding and beekeeping are potential areas of conflicts with carnivores. Thus,

large carnivore conservation in traditional farming landscapes does not only depend on the

biophysical environment, but also on the willingness of people to live with carnivores (Treves &

Karanth 2003; Dickman 2010).

Aims

As described above the overarching goal of this dissertation was to ‘identify system properties that

facilitate biodiversity conservation in traditional farming landscapes’ using birds and large carnivores in the

rapidly changing traditional farmland region of Southern Transylvania as a study system (Fig. 1.3).

In order to identify these properties this dissertation is divided into three sections. In Section A, I

focus on the effects of local and landscape scale land-use patterns on birds and large carnivores.

This Section also provides insights on the effects of potential land-use change on biodiversity. To

gauge the role of traditional land-use elements for biodiversity, in Section B, I focus on wood

pastures as one prominent example of such traditional elements. Here, we examined the structure

of wood pastures, and their use by woodpeckers and bears. Lastly, in section C, I use a social-

ecological systems approach to understand social drivers underlying human-bear coexistence. In

this part, we assessed the level of human-bear conflicts and the factors shaping the willingness of

people to live with bears. Thus, my specific aims were (Fig. 1.3):

A. How do local and landscape scale land-use and land-use change affect biodiversity (Chapters II-V)?

B. How do ancient wood pastures affect biodiversity (Chapters VI-VIII)?

C. How do links between the ecological system and the social system affect human-bear coexistence

(Chapters IV and IX)?

Chapter I

14

Figure 1.2 Examples of changes in land-use and land-use practices occurring in the study area. The pictures

show the change from relatively low intensity land-use to higher intensity land-use in arable land (a and b)

and in grassland (c and d). Intensification reduces the structural diversity of the landscapes through for

example reduced woody vegetation cover and land cover diversity. Pictures e and f show a change from

manual to mechanized hay harvest, reducing for example differences in sward heights throughout spring and

summer.

a) b)

c) d)

e) f)


15

Figure 1.3 Overview of the dissertation. Starting from the top, the social-ecological framework for

biodiversity conservation in traditional farming landscapes is presented, followed by the specific aims of this

dissertation. The middle part summarizes the empirical work of the dissertation divided into three sections,

each of these addressing a specific aim (A, B and C). The chapters contributing to the sections are indicated

by Roman numbers (I-IX). The synthesis of the dissertation, namely the system properties facilitating

biodiversity conservation and the conservation priorities for traditional farming landscapes, are presented at

the bottom.

Chapter I

16

Summary of included chapters

Chapters II-V in Section A address how land-use and land-use change affect biodiversity. In

Chapter II, we focused on the effects of land-use and land-use change on the passerine

community. We conducted repeated point-counts for breeding males in the three main land-use

types: arable fields, grasslands, and forests. While forest sites supported the highest bird species

richness (alpha diversity), arable fields and grasslands supported a larger number of different

species (beta-diversity) and communities were more heterogeneous than in forests. We used a snap-

shot natural experiment to gauge how anticipated agricultural intensification and land abandonment

would affect the bird community in farmland (arable fields and grasslands). Gradients in landscape

heterogeneity and woody vegetation cover measured at three spatial scales, ranging from local to

landscape-wide, were used to indicate anticipated changes in land-use and their effects on birds. We

also included the effects of topographic variables on birds because they represent key biophysical

gradients. Woody vegetation cover was the most important variable affecting the bird community.

Species composition changed along gradients of woody vegetation cover at all three spatial scales.

Total species richness asymptotically increased with local woody vegetation cover. In addition,

woody vegetation cover negatively affected open-country specialists at an intermediate scale, and

farmland birds at the landscape scale. Our study demonstrates the conservation value of traditional

farming landscapes for birds. To maintain bird diversity it is important to uphold gradients of

woody vegetation cover at multiple spatial scales, including open areas for open-country specialists.

Both the decrease of woody vegetation through land-use intensification and the increase of woody

vegetation through land abandonment, will negatively affect the farmland bird community as a

whole, although the impact may differ between different functional groups. Thus, the landscape

may benefit from different strategies aiming at mitigating either intensification or abandonment,

with their implementation depending on the area.

Chapter III focused on the effects agricultural intensification on an EU-protected

farmland bird species, the corncrake. Similarly to Chapter II we used changes in heterogeneity and

woody vegetation cover, measured at three spatial scales, as measures to reflect changes associated

with land-use intensification. In addition to heterogeneity and woody vegetation cover, we used

topographic and human disturbance variables to explain corncrake presence in the landscape. The

corncrake is typically considered a grassland species. However, we observed corncrakes in both

grasslands (e.g. hay meadows) and in the arable mosaic (e.g. alfalfa fields). Furthermore, corncrakes

were present more frequently in areas that were wet, flat and high in land cover diversity within a

scale of 100 ha. This study shows the importance of the heterogeneous land cover mosaic of

traditional farmland for the corncrake, where the loss of land cover diversity due to intensification

could severely reduce the availability of corncrake habitat. To assess this impact, we predicted the

distribution of suitable corncrake habitat under a scenario of land cover homogenization by

simulating a reduction in land cover diversity. Importantly, the loss of land cover diversity and

reductions of corncrake habitat were not linearly related, with even small losses in land cover


17

diversity resulting in a high loss of suitable corncrake habitat. Thus, pro-active conservation

measures for the corncrake should include the farmland mosaic beyond grasslands. Here it will be

important to encourage the persistence of mixed farming.

We examined the effects of land-use patterns on bear distribution in Chapter IV. Bear

distribution was indicated through a sign-based metric of bear activity, namely the proportion of

anthills destroyed by bears relative to the total number of anthills in a transect. We were also

interested in identifying specific hotspots for bears in the region, especially with regard to the

Natura 2000 area. We modelled bear activity in relation to anthropogenic, biophysical, and

connectivity variables, and based on this, predicted bear activity for the entire study area. Contrary

to our expectations bears were not influenced by distance to the nearest settlement. Instead,

connectivity to the Carpathian Mountains, where the source population resides, was the most

important variable explaining bear activity. This measure of connectivity was indicated through a

cost-distance metric, where the ‘cost’ for a bear to move through each possible land cover type was

scored by a local bear expert. In contrast, connectivity between forest patches contained within the

study area did not affect bear activity. Connectivity of the different forest patches was indicated

through ‘betweenness centrality’, which represents how well connected a forest patch is within the

forest patch network regardless of the other land covers between these patches. Furthermore, bear

activity was higher in more rugged areas with large forest blocks and low pasture cover. We did not

find particular hotspots of activity in the Natura 2000 area. Rather, predicted bear activity showed a

gradual increase toward the Carpathian Mountains, but was otherwise relatively homogenous

throughout the study. Our results suggest that conservation management for bears should primarily

maintain the connectivity to the Carpathian Mountains, which would require land-use management

beyond the Natura 2000 region. The lack of importance of connectivity between forest patches at

more local scales indicates that forest fragmentation has not yet reached a level that would affect

bears and that high connectivity remains throughout the study area. Therefore, emphasis should be

placed on preserving large connected forest blocks, especially in rugged areas where bears can find

shelter. In contrast to birds, reduced pasture cover and an increase in woody vegetation as result of

land abandonment is likely to positively affect bears.

Chapter V explored the role of large carnivores in a human dominated ecosystem by

analysing trophic interactions between forest mammals, including humans as apex predators. We

aimed at assessing the top-down effects of large carnivores on herbivores and mesopredators in

relation to direct and indirect human top-down effects and bottom-up effects. Wolves, bears,

domestic dogs (Canis familiaris), and humans represented apex predators, red deer (Cervus elaphus)

and roe deer (Capreolus capreolus) represented the herbivores, and the red fox (Vulpes vulpes)

represented the mesopredator. Bottom-up effects were indicated by pasture and forest cover. We

combined data on species encounter rates from camera traps and bear and wolf densities from

hunting records to assess trophic interactions with piecewise structural equation models. We found

that bears and wolves exert top-down control, especially on herbivores, and thus maintain their

Chapter I

18

ecological role in human-dominated landscapes. Nevertheless, direct and indirect human-top down

effects at multiple trophic levels affected species encounter rates more strongly. Furthermore,

herbivores were also limited by dogs brought into the system by humans. The importance of top-

down herbivore control was even more evident through the relatively weak effect of the land cover

(i.e. bottom-up) variables. On the other hand, through their limiting effect on predators, humans

may reduce the predators’ top-down control. Although not accounted for in this study, humans

may further affect trophic cascades by mediating bottom-up effects through landscape

modification. Thus, bears and wolves are important for the regulation of the ecosystem, but human

direct and indirect top-down effects are currently higher. The persistence of bears and wolves

should therefore be ensured in order to conserve the ecological character of these valued traditional

landscapes. We also believe that further understanding of the different human effects on trophic

interactions in modified landscapes should be a major research priority.

Chapters VI-VIII in Section B focus specifically on the conservation value of traditional wood

pastures to gauge the role of particular traditional land-use elements for biodiversity. To gain a

better understanding of wood pastures we first of all explored the characteristics, management, and

status of wood pastures in Chapter VI. Wood pastures were mainly dominated by oak and several

species of fruit trees, which differed from the tree community composition in forests. Wood

pastures also contained more ancient trees (trees of which the age can reach centuries) compared to

forest sites, but were relatively low in dead tree abundance and shrub cover. These characteristics

reflect the traditional management of wood pastures, which were created from forests by grazing

and selective tree removal. Remaining trees were not only valued for their shade, but oaks provided

timber and acorns for livestock, while fruit trees provided fruits. Specific characteristics of wood

pastures were determined by variables such as topography, distance to village, management, and

surrounding forest cover. For example, fruit trees were more prevalent close to villages, while many

dead trees were found in wood pastures surrounded by forests. A detailed study on the biodiversity

of one specific wood pasture showed the high potential of wood pastures to support high levels of

biodiversity. Worryingly, current management of wood pastures differed from traditional

techniques in several aspects, which could potentially threaten their persistence. First, most wood

pastures were grazed by sheep only, whereas mixed livestock grazing used to be the common

management type. Second, many large trees suffered from (anthropogenic) burning. Although

pasture clearing through controlled burning has been used for centuries, the current trend towards

uncontrolled burning is a major threat to wood pastures. Third, we found some evidence of (illegal)

tree cutting, while historically only branches were cut and the trunk remained (‘pollarding’). Wood

pastures are not consistently formally protected within the EU, and our results show that their

persistence in Transylvania is in danger. Therefore, a solid ecological understanding in combination

with knowledge on cultural and human livelihood importance of wood pastures needs to be

developed to further their recognition for formal conservation.


19

The role of wood pastures for biodiversity was explored in more detail in Chapters VII

and VIII. In Chapter VII we assessed the habitat value of wood pastures for an assemblage of six

woodpecker species. Woodpeckers are considered to be highly sensitive to changes in forest

management due to their large home ranges and their requirement of large trees and dead wood for

nesting and foraging. Since wood pastures retain elements of natural forests they may provide

additional habitat for the more forest-associated woodpeckers. Indeed, we found that species

richness in wood pastures was similar to forests, although species composition differed slightly.

Wood pastures were especially important for the green woodpecker (Picus viridis), which is

considered an open-country species, while forests were more important for the lesser-spotted

woodpecker (Dendrocopos minor), which typically avoids foraging in open areas. In contrast, the other

four woodpecker species occurred in both wood pastures and forests. Two of the protected species

were especially prevalent in wood pastures with a higher surrounding forest cover. Thus, wood

pastures provide valuable supplementary habitat for woodpeckers. Favourable characteristics of

wood pastures may be high food availability (e.g. ants and insects), nesting cavities in large trees,

and the provision of connectivity between different forest patches.

Chapter VIII showed that wood pastures also provide valuable supplementary habitat for

the brown bear. We found evidence for bear activity in 87% of the surveyed wood pastures

(indicated through destroyed anthills). Similarly to Chapter IV, bear activity was higher at a close

proximity to the Carpathian Mountains and in more rugged and forested areas. Bears may find

wood pastures suitable for foraging because of the availability of multiple food sources and the

shelter provided by woody vegetation. Grassland ant species are not found in forests, but form an

important source of protein for the mainly vegetarian bear. During autumn, wood pastures further

provide fruits and hard mast, although the use of these food sources by bears could not be assessed

within the time-frame of our study. These two studies show the high potential of wood pastures for

biodiversity conservation. We suggest to explicitly consider wood pastures in major EU

conservation policies, for example by stimulating the maintenance of scattered trees in extensively

managed pastures.

Section C takes a social-ecological systems approach to understand how links between the social

and ecological parts of the system affect human-bear coexistence. In Chapter IV we used

questionnaires to obtain an overview on human-bear conflicts in the study area, and correlated

perceived levels of conflicts with observed bear activity. Conflicts with bears occurred across the

study area. People reportedly suffered from damage to crops, orchards, and beehives, as well as

predation on livestock, while attacks on humans were rare. Cow herders had little problems with

bears. In contrast, about half of the shepherds suffered bear attacks on sheep during the past three

years. Interestingly, perceived level of damage to orchards and crops was positively correlated with

bear activity, while bear activity did not correlate to the number of sheep attacks or perceived level

of damage to beehives. These differences may be explained by differences in guarding

Chapter I

20

management. Orchards and crops are often left unguarded, whereas, sheep are actively guarded by

both shepherds and sheep guard dogs. The lack of a correlation between bear activity and

perceived damage to beehives may be more related to the low abundance of this technique among

participants. These results indicate the importance of local factors other than bear activity on the

prevalence of livestock predation and highlight the possibility for conflict mitigation (e.g. through

location of the sheep camp). We asked shepherds several additional questions to examine whether

livestock predation affected their attitudes towards bears. There was some tolerance towards bears,

despite occasional sheep predation. Shepherds suffering from a higher rate of bear attacks

nevertheless expressed a strong dislike of bears more frequently. About half of the shepherds were

unsupportive of immediate killing of bears after a sheep attack. Thus, the use of traditional sheep

herding techniques combined with the tolerance of some shepherds is likely to facilitate human-

bear coexistence in the region.

In Chapter IX, we combined questionnaires with semi-structured interviews for a more

holistic understanding of the social drivers underlying human-bear coexistence. The majority of

participants had a positive perception of coexistence. The questionnaires revealed general patterns

on social drivers underlying coexistence such as past negative interactions, perceived risks of

damage, attitude, and age. The interviews revealed three coexistence pathways highlighting the

causal mechanisms driving people’s willingness to coexist with bears. These pathways show

different ways in which ongoing interactions between the ecological system and the social system

shape the willingness of people to coexist with bears. The three pathways were defined by three

major themes, namely bears, humans, and management. The landscape had important mediating

effects on the pathways centred on bears and humans. For example, in the landscape-bear

coexistence pathway, people’s perceptions and beliefs about bears were largely shaped through

direct interactions and experiences with bears. The importance of direct interactions was further

emphasized in the landscape-human coexistence pathway where people’s tolerance towards bears

increased following positive encounters with bears, while livestock predation by bears decreased

people’s tolerance. Nevertheless, negative perceptions on coexistence were more likely to be

shaped by the perception of a high risk of potential conflicts with bears than by the actual

experience of damage caused by bears. Furthermore, the landscape-human-coexistence pathway

showed that genuine links between people and their environment (i.e. where people value their

natural surroundings) were important drivers of people’s positive attitudes towards bears and an

ascription of non-use values to bears. However, the management coexistence pathway revealed that

perceived inadequate management may erode the rural population’s tolerance for bears.

Management related to bear population management, conflict mitigation and compensation

payments, and trophy hunting was often perceived unsatisfactory by the participants. In addition,

the feeling of distrust towards management bodies and dis-empowerment further widened the gap

between management bodies and local stakeholders. We conclude that to avoid the escalation of

human-human conflicts over bears, where bears represent disagreements between local


21

stakeholders and management bodies, coexistence may be maintained or facilitated through: (i)

participation of local stakeholders to enhance the information flow and reduce distrust towards

management bodies; (ii) targeted education programs to address people’s specific beliefs and

concerns regarding bear-related issues; and (iii) the development and increased transparency of

current and alternative solutions for conflict mitigation.

Synthesis: System properties facilitating biodiversity conservation

This dissertation provides important insights on biodiversity drivers and patterns in traditional

farming landscapes. To start with, it demonstrates the large biodiversity value of traditional farming

landscapes in general, and calls for an increased recognition of these systems for biodiversity

conservation. This is especially urgent since anticipated land-use change and the loss of traditional

farming practices may cause significant biodiversity declines in traditional farming landscapes.

Undoubtedly, there are many system properties of traditional farming landscapes that potentially

facilitate biodiversity conservation. Some of these are beyond the focus of this dissertation such as

traditional livestock and crop rotation schemes, diverse crop systems, low agro-chemical input, and

land-use that is not optimized to produce maximum yields. However, through the consideration of

two different animal taxa, this dissertation reveals six important system properties that support high

biodiversity in Transylvania’s traditional farming landscape.

1. Similar proportions of main land-use types

Biodiversity was supported by the heterogeneous character of the traditional farming landscape at

multiple spatial scales. At the scale of the study area, relatively similar proportions of the three main

land-use types likely support high biodiversity. Habitat loss and fragmentation are considered major

drivers of mammal and bird declines (Andrén 1994; Monastersky 2014). However, fragmentation

effects usually become visible below a threshold of 30% of available habitat (Andrén 1994; Hanski

2011), which is close to the proportional cover of the study area’s three main land-use types. The

approximately one-third of forest cover also seemed to provide sufficient habitat connectivity for

the brown bear (Chapter IV). Habitat connectivity at large scales is an important system property as

it can facilitate species’ movements and dispersal in fragmented landscapes and maintain gene flow

and metapopulation dynamics (Fischer & Lindenmayer 2007; Kopatz et al. 2012). For the brown

bear in Southern Transylvania, the existing connectivity between the study area and the source

population in the Carpathian Mountains, provided by configuration and composition of land

covers, was found to be particularly important (Chapter IV). In other parts of the brown bear’s

European range, in contrast, forest connectivity is degraded to a degree that limits the species’

expansion (Fernández et al. 2012).

Within farmland, bird species composition was not determined by land-use per se (e.g.

arable land vs grassland), but was influenced by different environmental gradients (Chapter II). The

availability of both grassland and arable land in relatively large proportions may support the similar

Chapter I

22

species composition of birds (Chapter II) and butterflies (Appendix I) in both land-use types, for

example through constant spill-over (Tscharntke et al. 2012). These processes may prevent the

divergence of distinct grassland and arable land communities but maintain more diverse

communities at the landscape scale. Although plant species composition differed between

grasslands and arable land, a substantial number of species was shared and both land-use types

were important contributors to the total species pool (Appendix II). In addition, a forest-farmland

mosaic facilitates spill-over effects from forests to farmland (Tscharntke et al. 2012), and we

observed a considerable number of forest bird species in farmland (Chapter II). Thus, the observed

proportions of land-use types in the study area support species associated with farmland as well as

with forests, with all three major land-use types contributing to high regional biodiversity.

2. Complementary or supplementary habitat

Heterogeneous landscapes can further support biodiversity through complementation and

supplementation of habitat at the landscape scale (Dunning et al. 1992). Landscape

complementation is provided in landscapes in which species encounter all required spatially

separated habitats containing necessary resources, while landscape supplementation is provided in

landscapes in which species encounter additional habitats that contain similar resources (Dunning

et al. 1992). We observed several species in land-use types outside their core habitat. For example,

wood-pastures were extensively used by different woodpecker species and the brown bear

(Chapters VII and VIII). The retention of forest structures across the landscape in traditional

farming landscapes thus provides supplementary if not complementary habitat for forest species

(Mikusinski & Angelstam 1998). Similarly, the corncrake was present throughout the arable mosaic

despite being considered a grassland species (Chapter III). Uncropped arable land in combination

with field margins or ditches may be important in providing resources similar to grasslands, such as

safe breeding and sheltering sites and high insect availability (Corbett & Hudson 2010; Budka &

Osiejuk 2013; Josefsson et al. 2013).

3. High woody vegetation cover and heterogeneity

At a more specific level, traditional farming landscapes supported biodiversity through the presence

of gradients in woody vegetation cover, including semi-natural vegetation, and through

heterogeneity, measured as the composition and configuration of land cover. High woody

vegetation cover supported bird (Chapter III), butterfly (Appendix 1) and plant (Appendix 2)

species richness locally, possibly by providing a range of resources such as refuge areas, nesting,

sheltering, and foraging sites (e.g. Benton et al. 2003; Ernoult & Alard 2011), or by facilitating

cross-habitat movements and spill-over (Tscharntke et al. 2012). Importantly, not all taxa respond

linearly to woody vegetation cover or heterogeneity. For example, bird richness increased

asymptotically with woody vegetation cover, which was especially evident in grasslands (Chapter II)

which harbours a large number of open-country species that disappear beyond certain levels of


23

woody vegetation cover (Sanderson et al. 2013). These findings demonstrate the need to also

maintain relatively homogenous, open areas (Batary et al. 2011b). Another case of important non-

linearity was revealed by the simulated severe reduction of corncrake habitat at relatively modest

levels of land cover homogenization (Chapter III).

We also identified the significance of the landscape context for effects of woody vegetation

cover and heterogeneity. Woody vegetation cover and heterogeneity influenced different species or

groups of species of birds (Chapter II), butterflies (Appendix 1), and plants (Appendix 2) at larger

spatial scales, and sometimes with effects opposite to those at smaller scales. Furthermore, we

found biodiversity to be affected not only by processes at multiple spatial scales, but effects also

differed between species, which may depend on their specific resource needs (Lindenmayer &

Fischer 2006). To conclude, the availability of woody vegetation cover and heterogeneity at

different spatial scales are important drivers of biodiversity in traditional farming landscapes.

Moreover, we demonstrated the need to further understand the scale dependence of different

species and across different taxa.

4. Traditional land-use practices

The presence of high woody vegetation cover and heterogeneity is linked to traditional semi-

subsistence farming practices. Such farming, including high degree of manual labour and few agro-

chemical inputs, is thus key to maintaining biodiversity. The manual cutting of hay in a mosaic

pattern, for example, provides a variety of sward heights throughout the breeding season of

corncrakes, thereby facilitating their presence in agricultural land (Chapter III). Wood pastures were

created by traditional silvo-pastoral practices but current management techniques differ from the

traditional ones and may severely threaten the persistence of wood pastures in the landscape

(Chapter VI). This change in management is likely to have a negative effect on the biodiversity

supported by wood pastures. As another example, the use of traditional livestock husbandry

techniques allowed coexistence between humans and large carnivores (Chapter IV and IX). The

combination of shepherds, livestock guard dogs, and nightly confinement of livestock are

successful in reducing livestock conflicts worldwide (Rigg 2001; Gehring et al. 2010), as well as in

Romania, as demonstrated here. This provides an important insight for European regions into

which large carnivores return after their earlier extirpation, but in which the loss of these husbandry

techniques and resulting conflicts hamper their successful establishment (Enserink & Vogel 2006;

Chapron et al. 2014).

5. Top-down carnivore regulation

The presence of large carnivores in the landscape can benefit biodiversity through top-down

control on mesopredators and herbivores (Estes et al. 2011; Ripple et al. 2014), which induce

trophic cascades that affect species at multiple trophic levels (Letnic et al. 2009). The importance of

trophic cascades for biodiversity have mainly been observed in wilderness areas (Ripple & Beschta

Chapter I

24

2012b), whereas the role of large carnivores in structuring human-dominated ecosystems remains

unclear (Sergio et al. 2014). We found limited evidence for top-down control on a mesopredator

(the red fox; Chapter V), which may be explained by large differences in body size (Donadio &

Buskirk 2006; Ritchie & Johnson 2009), or by densities of wolves and bears too low to effectively

limit foxes. In contrast, we found top-down limitation of wolves and bears on herbivores (Chapter

V), which indicates the importance of their persistence for the ecosystem (for example by limiting

overgrazing, enhancing vegetation growth, and maintaining biodiversity in general; Terborgh et al.

2001; Estes et al. 2011). Still, to fully understand the role of carnivores in traditional farming

landscapes future research should aim to understand (i) the extent and effects of possible trophic

cascades induced by carnivore top-down herbivore control; and (ii) the role of human bottom-up

and top-down effects on trophic cascades in combination with the effects induced by carnivores.

6. Human-nature connections

Large carnivore persistence does not only depend on the biophysical environment, but also on the

degree to which the rural population is willing to coexist with large carnivores (Treves & Karanth

2003). People in our study area had a general positive perception of human-bear coexistence

(Chapter IX). Their ability to tolerate carnivores partly stemmed from genuine links between

humans and nature, with people valuing the natural heritage. People’s culture shapes the landscape,

but the landscape in turn also shapes the culture of people (Pretty 2011). The centuries of co-

occurrence of humans and bears in Southern Transylvania probably shaped human culture to

accept and adapt to living with carnivores (see also Glikman et al. 2012). These strong ties between

people and the landscape possibly form the core of people’s values and sustainable use of natural

resources. Disconnecting these bonds could potentially destroy people’s culture and with it the

associated natural heritage (Pretty 2011). In the case of large carnivores, for example, perceived

failures of top-down managing institutions may harm carnivores through reduced tolerance and

increased poaching (Chapter IX; Bell et al. 2007; Liberg et al. 2012; Gangaas et al. 2013).

Conservation priorities for traditional farming landscapes

Traditional farming landscapes have a high conservation value, and facilitating biodiversity

conservation in these landscapes should be of priority worldwide. The maintenance or preservation

of the six system properties highlighted in this dissertation should be a central part of conservation

measures targeting traditional farming landscapes. Biodiversity was greatly supported through the

heterogeneous character of the forest-farmland mosaic. This heterogeneous character, however, is

tightly linked to the historical multi-functionality of the landscape and the practice of small-scale

semi-subsistence farming. Yet, it is these features that will vanish under anticipated land-use change

in form of agricultural intensification or land abandonment. Agricultural intensification will cause

major declines in biodiversity. Although land abandonment may benefit certain forest species (e.g.


25

bears and forest birds), the associated decline of farmland species will significantly reduce the

overall species pool. Moreover, we found that humans and bears can coexist in the landscape at

current conditions, without the need to set additional land aside for carnivore conservation. Thus,

conservation priorities should integrate farming and biodiversity conservation through encouraging

farming that maintain the heterogeneous character of the landscape (i.e. land sharing; Fischer et al.

2008).

Conservation initiatives that prevent biodiversity loss in heterogeneous farmland are

considered to be more (cost-) effective than restoration initiatives after biodiversity declines driven

by land-use change (Kleijn et al. 2011). Yet, current conservation measures in the EU are poorly

adapted to support biodiversity in heterogeneous traditional farming landscapes, which face very

different conservation challenges from more intensified landscapes (Tryjanowski et al. 2011;

Mikulcak et al. 2013; Sutcliffe et al. 2014). The two main measures to mitigate farmland biodiversity

loss in Europe are agri-environment schemes (AES) of the Common Agricultural Policy and the

Natura 2000 network. The effectiveness of AES for farmland biodiversity conservation has been

questioned because biodiversity continues to decline (Kleijn et al. 2006; Kleijn et al. 2011; Pe'er et

al. 2014). Through the voluntary uptake of AES by farmers, implementation of AES is often

restricted to small field or farm scales and may be one of the reasons why AES can be unsuccessful

(Whittingham et al. 2007). Furthermore, local-scale AES are generally considered to be ineffective

in highly heterogeneous landscapes (Tscharntke et al. 2005; Concepción et al. 2008; Batary et al.

2011a). The Natura 2000 network, on the other hand, designates sites at larger scales that protect

specific habitats and species listed under the Habitats and Birds Directives, and can have positive

biodiversity impacts (Gruber et al. 2012; Brodier et al. 2013; Pellissier et al. 2013). Nonetheless,

these measures also may be limited in their effectiveness, due to for example missing cooperation

between EU member states and a lack of trans- and interdisciplinary research to obtain a holistic

social-ecological understanding of the selected sites (Apostolopoulou & Pantis 2009; Albuquerque

et al. 2013; Popescu et al. 2014). In our study area, the implementation of the management plan for

the Natura 2000 area is pending, and it remains unclear how successful the forthcoming plan will

be in maintaining the heterogeneous character of the landscape. Furthermore, the brown bear

(listed under the Habitats Directive) depends on connectivity to the Carpathians Mountains outside

the Natura 2000 area, which is threatened by Romania’s increasing deforestation rates (Knorn et al.

2012; Griffiths et al. 2013), also in places where the Natura 2000 management plan will not apply.

Thus, there is a need to develop more holistic visions for biodiversity conservation in this and

other traditional farming landscapes, which include areas inside and outside protected areas.

Building upon the identified system properties such a holistic approach may firstly

comprise ‘broad and shallow’ landscape-scale conservation strategies targeting the entire forest-

farmland mosaic. Secondly, large scale conservation measures may be complemented with more

‘deep and narrow’ conservation measures targeting specific species, land-use types, threats, or

Chapter I

26

traditional practices. Third, conservation initiatives should encourage the integration of the entire

social-ecological system by recognizing important links between people and the environment.

The application of AES at the landscape scale has already proven beneficial for a range of

species (Merckx et al. 2009; Dallimer et al. 2010). In the case of traditional farming landscapes,

large-scale conservation measurements could provide the more ‘broad and shallow’ landscape

perspective to maintain the heterogeneous landscape character at multiple spatial scales

(Concepción et al. 2008). These measures could focus on maintaining similar proportions of land-

use types, large-scale habitat connectivity, complementary and supplementary habitat of different

species, and the gradients in land cover heterogeneity and woody vegetation cover. Applying this

landscape perspective may, for example, prevent the under-valuation of certain habitats (e.g. wood

pastures) and allow the inclusion of conservation strategies for supplementary species habitat (e.g.

arable land for the corncrake and butterflies).

‘Deep and narrow’ conservation measures need to complement landscape-scale

conservation measures to ensure the persistence of specific species (Koleček et al. 2014). Here, it

seems vital to advance our understanding of scale-dependent responses of different species to

woody vegetation cover and land cover heterogeneity (Pickett & Siriwardena 2011; Tscharntke et

al. 2012). Biodiversity responses in our study also differed between land-use types indicating the

need for distinct and more detailed strategies for grasslands and arable fields. Similarly, biodiversity

may benefit most from policies targeted to mitigate either abandonment or intensification. For

example, abandonment may be more prevalent in remote grasslands on steep slopes and could be

mitigated by providing incentives to maintain rotational livestock grazing to remove shrubs. In

contrast, intensification is more likely to occur in accessible arable land where the retention of

woody vegetation cover should be prioritized. Lastly, these conservation measures could focus on

the preservation of key elements of traditional land-use and practices (e.g. livestock herding

techniques) and on strategies for their future integration in new land-use systems (Plieninger et al.

2006).

The above-mentioned priorities aim to support biodiversity conservation. However,

ultimately, the persistence of biodiversity in traditional farming landscapes will depend on

navigating social-ecological change so that it does not only maintain biodiversity but also benefit

local people. Traditional farming practices have become largely unviable, and standard conservation

policies take a ‘preservation approach’ where financial incentives are provided for people to

maintain traditional farming practices (Fischer et al. 2012; Plieninger & Bieling 2013). These

strategies, however, may only be successful in the short term, because human-environment bonds

are not primarily driven by economic incentives (Milcu et al. 2014). In our study, people did not

tolerate bears for their economic benefits but because of non-use values people ascribed to bears.

Understanding and addressing these values and links between people and the environment could

reduce conflicts between rural populations and conservation initiatives (Ives & Kendal 2014).

Moreover, the links between people, their activities, and the environment could form the base for a


27

more integrated ‘transformation approach’ (Fischer et al. 2012). Such an approach should foster

new links between the social and the ecological parts of the system aimed at maintaining the system

properties of traditional farming landscapes that support biodiversity but also fosters human-well-

being. Alternative development pathways could be created through, for example, the uptake of

agro-ecological and organic farming or the development of agro-ecotourism (Hole et al. 2005;

Young et al. 2010). Because people may have aspirations for the future different from those

prioritized by conservation (Milcu et al. 2014), community participation and the support of bottom-

up driven initiatives are essential for the development of holistic conservation strategies.

Importantly, biodiversity conservation of traditional farming landscapes can only be facilitated if

initiatives to support other capitals (e.g. social, human, financial, physical) are developed

simultaneously (Mikulcak et al. 2015). In the end, the future of traditional farming landscapes in a

globalized world will depend on how well people can capitalize on the available opportunities

(Hanspach et al. 2014), and thus successful biodiversity conservation will hinge on the integration

of the entire social-ecological system.

Conclusion

The unique natural heritage of traditional farming landscapes is under severe pressure from the

regional impacts of globalization, with land-use intensification and land abandonment threatening

to erode regional biodiversity. The future of traditional farming landscapes and its associated

biodiversity depends on the navigation route of the social-ecological system through these

profound changes. Using the example of birds and large carnivores in the changing traditional

farming region of Southern Transylvania, this dissertation revealed six important system properties

that facilitate biodiversity conservation. Drawing on the identified properties this dissertation

highlighted several conservation priorities that can help to maintain biodiversity in traditional

farming landscapes. There is a clear need to develop a conservation vision for traditional farming

landscapes that takes a holistic approach and complements ‘broad and shallow’ landscape-scale

conservation of the entire forest-farmland mosaic with more ‘deep and narrow’ conservation

initiatives targeting specific species, land-use types, threats, or traditional practices. More

importantly, integration of the entire social-ecological system within this vision is necessary to

maintain biodiversity and simultaneously foster the well-being of the rural population whose

livelihoods are strongly interconnected with the natural environment. There is an urgent need to

develop visions that navigate traditional farming landscapes through ongoing land-use change

because these landscapes are rapidly disappearing worldwide. Ensuring the persistence of

traditional farming landscapes in the future would contribute considerably to the challenging yet

crucial goal to halt global biodiversity decline.

28

29

Section A: Land-use

effects on biodiversity

This Section includes Chapters II-IV and examines the effects of local and landscape scale land-use

patterns on birds and large carnivores, and how future land-use change may affect them. Chapter II

aims to gauge the likely effects of future land-use change on the passerine community, with

emphasis on land-use intensification and land abandonment. Chapter III aims to assess the

sensitivity of an EU-protected farmland bird species, the corncrake, to possible future land-use

intensification. Chapter IV examines the effects of land-use patterns on brown bear distribution.

Chapter V explores the role of large carnivores in the ecosystem by analyzing the top-down effects

of large carnivores on herbivores and mesopredators in relation to direct and indirect human top-

down effects and land-use bottom-up effects.

30

31

Chapter II

32

33

Chapter II

Maintaining bird diversity in a traditional farming landscape

in Eastern Europe

Ine Dorresteijn, Jacqueline Loos, Jan Hanspach, Cosmin Ioan Moga, Alin David,

Lunja Marlie Ernst, Joern Fischer

Manuscript

‘Farmland birds are at their lowest levels since records began.’ – BBC News 2014

‘We know that wildlife is bleeding from the countryside, we know the reasons behind it and we know how

to fix it. We just aren’t fixing it because government isn’t bothered. No news there then.’ – Mark Avery

Chapter II

34

Bird diversity in traditional farming landscapes

35

Abstract

Traditional farming landscapes harbor high biodiversity worldwide. However, farmland biodiversity

is increasingly threatened by agricultural intensification and land abandonment. We aimed to assess

the drivers of biodiversity in a traditional farming landscape in order to gauge the likely effects of

agricultural intensification and land abandonment on bird communities. We examined the

responses of species composition, species richness, and richness within functional groups, to

woody vegetation cover, land cover heterogeneity, and topography (each measured at local,

context, and landscape scales). We conducted repeated point counts for breeding birds in a

randomly selected set of 30 forest, 60 grassland, and 60 arable sites, which spanned stratified

gradients in heterogeneity and woody vegetation cover. Species composition in forests differed

from that in grassland and arable land. Species composition in grassland and arable land responded

to gradients of woody vegetation cover at all three scales. Within grassland and arable land, total

species richness, richness of farmland birds, and of forest specialists, asymptotically increased with

local-scale woody vegetation cover. In contrast, richness of open-country species responded

negatively to context-scale woody vegetation cover, and other farmland birds responded negatively

to landscape-scale woody vegetation cover. Our results show that different groups of birds would

be impacted differently by contrasting future land-use change scenarios. The regional bird

community as a whole would benefit from the maintenance of gradients in woody vegetation cover

across multiple scales.

Chapter II

36

Introduction

Agricultural expansion and intensification have caused biodiversity decline worldwide (Tilman et al.

2001; Foley et al. 2005), with the widespread loss of landscape heterogeneity identified as one of

the major drivers (Benton et al. 2003). Agricultural land holds a large part of the world’s

biodiversity (Pimentel et al. 1992) and occupies 40% of Earth’s land surface (Foley et al. 2005).

Growing recognition of the value of farmland biodiversity has sparked major national and

international conservation initiatives, but nevertheless, farmland biodiversity continues to decline

(Kleijn et al. 2011). In contrast, low-intensity land-use systems, such as traditional farming

landscapes, often harbor exceptional biodiversity and are therefore of great conservation

significance (Tscharntke et al. 2005; Kleijn et al. 2009).

Traditional farming landscapes occur in regions where farming practices have changed

little over long periods of time, and are typically characterized by large amounts of semi-natural

vegetation and high heterogeneity in land covers and structural elements (Plieninger et al. 2006).

The value of traditional farming landscapes for biodiversity has been noted worldwide

(Ranganathan et al. 2008; Takeuchi 2010; Robson & Berkes 2011; Liu et al. 2013), including in parts

of Eastern Europe (Fischer et al. 2012). However, traditional farming landscapes have come under

severe pressure from land-use intensification, and in economically marginal areas, from land

abandonment (Lepers et al. 2005; Plieninger et al. 2006). Land-use intensification influences

landscape structure via the loss of non-crop landscape elements (including semi-natural vegetation,

shrub areas, and woodlands) and consequently, homogenization of the landscape (Benton et al.

2003). Land abandonment typically affects landscape structure by transforming agricultural land

into shrubland, which may eventually turn into forest (Rudel et al. 2005; Kuemmerle et al. 2008).

To effectively conserve biodiversity in traditional farmland facing potential land-use changes thus

requires, first of all, a solid understanding of how different species respond to woody vegetation

cover and landscape heterogeneity.

Here, we present a comprehensive, regional-scale case study on the drivers of bird diversity

in a traditional farming landscape that is subject to land-use change. Our study investigated bird

community responses to landscape structure in Southern Transylvania, Romania, and was

specifically designed to cover the entire agricultural mosaic, spanning large gradients in

heterogeneity and woody vegetation cover. Transylvania’s farmland has changed relatively little

since pre-industrial times, and traditional semi-subsistence farming has maintained a species-rich

mosaic of arable fields, grasslands, and forests (Wilkie 2001; Cremene et al. 2005; Akeroyd & Page

2007). In recognition of its biodiversity, Europe’s largest lowland protected area within the Natura

2000 network has been established in Southern Transylvania to conserve its farmland biodiversity.

Yet, the region is undergoing rapid socio-demographic and land-use changes, which may

significantly impact biodiversity in the future (Mikulcak et al. 2013). Traditional farming has

become economically unviable, causing land abandonment by the rural population in some cases,

and agricultural land-use intensification in others (Hanspach et al. 2014).


37

We used a snapshot ‘natural experiment’ that substituted space for time (Diamond 1986)

to gauge the likely effects of future land-use change on Transylvania’s bird community. Farmland is

already intensified in parts of Transylvania, especially in the broadest valleys, whereas in remote or

rugged areas, farmland is increasingly being abandoned. We used this opportunity of impending

land-use change to strategically select survey sites along the full existing gradients of tree and shrub

cover (from open landscapes to forests) and landscape heterogeneity (from low to high

heterogeneity). Specifically, we assessed the responses of (1) bird community composition, and (2)

overall species richness and richness of species with different habitat specializations, to gradients in

woody vegetation cover, landscape heterogeneity, and topography, each measured at three different

spatial scales.

Methods

Study area

The study area covered 7441 km2 in the foothills (230 m to 1100 m above sea level) of the

Carpathian Mountains in Southern Transylvania, Romania (Fig. 2.1a). The region contained 28%

forest, 24% pasture, and 37% arable land. The remaining land cover included villages (typically

<1000 inhabitants), towns, water bodies, and permanent crops. Forests were dominated by

hornbeam (Carpinus betulus), oak (Quercus sp.), and beech (Fagus sylvatica). Most agriculture occured at

low intensities and small scales, with some exceptions, especially in the broader valleys. The major

crops were maize, alfalfa and wheat. Pastures and hay meadows occurred on the slopes and were

grazed by sheep, goats and cattle.

The landscape had an overall heterogeneous character, because of small farm sizes (often

<1 ha), a patchwork of small fields and field boundaries, and the occurrence woody vegetation

throughout fields and grasslands. Woody vegetation occurred in linear features such as lines of

trees and streamside vegetation, but also in small patches of shrubs and trees, small forest stands,

and scattered shrubs or trees. The most common trees in farmland were similar to those in forest,

but fruit trees (e.g. Prunus sp.) and Robinia pseudoacacia were also common. Common farmland shrub

species were Salix sp., Crataegus monogyna, Prunus sp., Corylus avellana, and Rosa sp. Almost half of the

study area was contained within Natura 2000 sites, including Sites of Community Importance (SCI,

Habitats Directive) and Special Protection Areas (SPA, Birds Directive; Fig. 2.1b).

Study design

Our design followed the principles of a snapshot ‘natural experiment’ in that we used randomly

selected and replicated experimental units within pre-defined strata of ecologically relevant

landscape gradients (Diamond 1986). We considered two spatial scales for stratification, namely

village catchments (defined below) and survey sites. We selected 30 village catchments and within

each selected five survey sites (Fig. 2.1b & 2.1c). Of a total of 150 sites, 30 were located in forest,

Chapter II

38

60 in grassland, and 60 in arable land. We collectively refer to grassland and arable land as

‘farmland’.

Figure 2.1 Study area. (a) Romania with our study area highlighted, (b) our study area with selected village

catchments highlighted, and (c) example of one of the village catchments showing different types of land

cover and five survey sites

Village catchments were chosen for stratification at the landscape scale because they

constitute both ecologically and socially meaningful units (Angelstam et al. 2003). We delineated

village catchments using a cost-distance algorithm implemented in ArcGIS, which allocated each

pixel to the village with the lowest travel cost (indicated by slope) to this pixel. These

topographically based village catchments closely reflected historical land-use responsibilities, and

were more suitable than official administrative boundaries, which were available only at the

commune level (with communes typically comprised of 3-5 villages). Of 448 village catchments, we

randomly selected 30, stratified to cover full gradients in terrain ruggedness (standard deviation of

the altitude; low, medium, high; defined by the upper, middle, and lower terciles) and protection

level (SCI, SPA, no protection; see Table S2.1 in Supplementary Material).

Sites were also randomly selected. For this, we used the Corine Land Cover Digital map

2006 (CLC) to identify three main land-uses, namely forest, grassland and arable land. Sites were

circular areas of one ha (i.e. radius of 56 m). To ensure that the entire site was located within a

single land-use cover, we buffered all edges between land-use covers and all roads by 60 m. These

areas were then masked prior to the random selection of sites. Furthermore, we only selected sites

within an area of 400 m from the closest track to ensure that all five sites within a village could be


39

surveyed within one morning. Notably, arable sites included fields as well as fallow land and field

margins. The median distance between neighboring sites was 606 m (the minimum was 200 m).

Grassland and arable sites were cross-stratified to cover full gradients in heterogeneity

(low, medium, high; defined by terciles) and percent woody vegetation cover (low (0-5%), medium

(5-15%), and high (15-60%)). Within our sites, heterogeneity and woody vegetation cover were

independent variables, and all possible cross-combinations were replicated, except for sites with

low heterogeneity and high woody vegetation cover, which did not exist (see Table S2.2).

Heterogeneity was measured as the standard deviation of 2.5 m monochromatic SPOT 5 satellite

data (©CNES 2007, Distribution Spot Image SA). Percent woody vegetation cover was derived

from a supervised classification of the 10 m panchromatic channels of SPOT 5 data (©CNES

2007, Distribution Spot Image SA) using a support vector machine algorithm (Knorn et al. 2009).

The resulting high-resolution map of woody vegetation cover showed the vast majority of large

shrubs, scattered trees, and linear features within the study area.

Environmental variables

We calculated environmental variables at three scales. The local scale covered the site (one ha), and

approximately corresponded to the typical home range size of breeding passerines (Cramp 2000).

The context scale covered an area of 50 ha (i.e. radius of 400 m), and was used because birds

responded to landscape metrics at this scale elsewhere (Barbaro & Van Halder 2009). The

landscape scale corresponded to the village catchment (mean ± SD: 2046 ± 1123 ha).

Within each scale we chose at least one variable for each of the following features: woody

vegetation cover, landscape heterogeneity, and topography (Table 2.1). Local variables included the

proportion of woody vegetation cover (range: 0-59%), remotely sensed compositional

heterogeneity, and a topographic wetness index. Local heterogeneity and woody vegetation cover

were calculated as described above. The topographic wetness index was calculated as a function of

slope and topographic position (Fischer et al. 2009).

Context variables included the proportion of woody vegetation cover (range: 0-57%),

Simpson’s diversity index of land cover (hereafter context land cover diversity), and terrain

ruggedness. Simpson’s diversity index of land cover was calculated using FRAGSTATS 4.2

(McGarigal et al. 2012), based on a raster of all land-use type included in the CLC map (i.e.

including land-uses not considered for site selection). Terrain ruggedness was calculated as the

standard deviation of altitude (based on the Advanced Spaceborne Thermal Emission and

Reflection Radiometer (ASTER) Global Digital Elevation Model Version 2 (GDEM V2)).

Landscape variables again included the proportion of woody vegetation cover (range: 9-

74%), landscape edge density, and Simpson’s diversity index of land cover, proportion of pasture

cover, and terrain ruggedness. The proportion of woody vegetation at the landscape scale was

primarily driven by the amount of forest cover, and we therefore used percent forest cover as

indicated by the CLC. Simpson’s diversity index of land cover and terrain ruggedness were

Chapter II

40

calculated as described above. Edge density (in m ha-1) is a measure of configurational

heterogeneity which standardizes edge per unit area; this meant it was comparable between village

catchments of different sizes. Edge density was calculated using FRAGSTATS 4.2, based on all

land-use types in the CLC map.

Table 2.1 Overview of environmental variables used for generalized linear modelling.

Scale Variable Description

Local

(LC)

Woody vegetation cover

(WVC)

Percent woody vegetation cover derived from a supervised

classification of the panchromatic channels of SPOT 5 Satellite

dataa

Heterogeneity (HET) Remotely sensed compositional heterogeneity indicated by the

reflectance of land surfaces with a resolution of 2.5 m x 2.5 m

measured using the monochromatic channel of SPOT 5 dataa.

Topographic wetness

index (TWI)

Measure of soil wetness, calculated as a function of slope and

topographic positionb

Context

(CT)


(WVC)

Percent woody vegetation cover derived from a supervised

classification of the panchromatic channels of SPOT 5 dataa

Land cover diversity

(LCD)

Simpson’s diversity index of land cover, based on a raster of all

land-use types derived from CLCc,d

Terrain Ruggedness (TR) Terrain ruggedness calculated as the standard deviation of the

altitudeb

Landscape

(LS)


(WVC)

Percent cover of forest derived from CLCc

Land cover diversity

(LCD)

Simpson’s diversity index of land cover, based on a raster of all

land-use types derived from CLCc,d

Edge density (ED) Edge density of land cover (in m ha-1) based on a raster of land-

use types derived from CLCc,d

Pasture cover (PC) Percent cover of pasture derived from CLCc

Terrain Ruggedness (TR) Terrain ruggedness calculated as the standard deviation of the

altitudeb a©CNES 2007, Distribution Spot Image SA bBased on the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital

Elevation Model Version 2 (GDEM V2) with a spatial resolution of 30 m x 30 m cCLC: Corine Land Cover Digital map 2006 dCalculated using FRAGSTATS 4.2 (McGarigal et al. 2012)


41

Bird surveys

Each site was surveyed three times during the breeding season by one of four experienced

observers, using 10 minute point counts based on visual and auditory observations (Bibby et al.

2000). Based on a pilot study, three surveys proved sufficient to representatively capture the species

within a site (Loos et al. 2014). Only singing males were recorded and included in the analysis –

that is, we excluded non-singing birds, corvids, raptors and aerial foragers. Surveys were carried out

in suitable weather conditions between 05:30 AM and 11:00 AM, between mid-April and mid-July

2012.

Statistical analysis

We analyzed the relationship between environmental variables and species presence, pooled across

the three repeats, using multivariate methods and regression modelling. We first compared species

richness using Analysis of Variance (ANOVA), and assessed patterns of beta diversity using species

accumulation curves (Thompson & Withers 2003), between the three main land-use types (forest,

grassland, arable land).

Second, we used detrended correspondence analysis (DCA) to compare bird community

composition between the three land-use types, and visualized the relationship of species

composition and environmental variables for farmland sites. DCAs were performed on bird species

occurring in more than one site, and rare species were downweighted (Oksanen et al. 2013).

Because forest communities strongly differed from farmland communities, further analyses were

performed on farmland sites only.

Third, we modeled total species richness and species richness of forest specialists,

farmland birds (defined here as farmland birds using shrubs and trees) and open-country birds, as

response variables within generalized linear mixed effects models with Poisson error distribution.

The models included the environmental variables and the quadratic term of local woody vegetation

cover as explanatory variables (fixed effects), and village catchment and site-level (to account for

overdispersion) as nested random effects. We also included the interaction terms of local woody

vegetation cover and local heterogeneity with land-use type (arable versus grassland), because the

effects of heterogeneity and woody vegetation cover may differ between arable land and grassland.

Prior to modeling, we log-transformed local and context woody vegetation cover and local

heterogeneity; confirmed that variables were not strongly correlated (ρ <0.6); and standardized all

variables to zero mean and unit variance. The final models were obtained through a stepwise

backward selection (p ≤0.01) with model comparisons using Chi-square statistics (Zuur 2009), and

did not exhibit spatial autocorrelation in the residuals. To visualize the effect of local woody

vegetation cover, we plotted predicted total species richness (considering only fixed effects) as a

function of local woody vegetation cover between 0% and 60% cover (the observed range within

the farmland sites) for arable and grassland sites separately. All analyses were implemented in the

R-environment using the packages ‘vegan’ and ‘lme4’ (R Development Core Team 2013). Habitat

Chapter II

42

specialization was assigned to all bird species by two Romanian experts (CIM and AD), based on

Birds of the Western Palearctic (Cramp 2000) and specific literature from Romania (Linţia 1954,

1955; Ciochia 1992).

Results

Bird diversity

We identified 61 breeding bird species (see Table S2.3). Species richness was higher in forests

(mean ± SE: 10.6 ± 0.53) than grasslands (5.5 ± 0.35) and arable land (5.5 ± 0.52; F2,147 = 28.18,

p<0.001). Despite having the highest alpha diversity, forests had lower beta diversity than arable

and grassland sites (Fig. 2.2).

Figure 2.2 Bird species accumulation curves for forest, grassland and arable sites. Ninety-five percent

confidence intervals are indicated by grey shading.

Community composition

Species composition in forest sites differed from grassland and arable sites, and forest sites were

more homogenous in species composition than grassland and arable sites (Fig. 2.3a). Bird

community composition strongly overlapped in grasslands and arable sites, but exhibited clear

gradients (both axes approximately covered one species turnover; length of gradient first DCA axis

= 5.34, second DCA axis = 3.82; Fig. 2.3b, see Fig. S2.1 for species scores of the ordination). The

gradients of bird composition in farmland were related to environmental variables at all three

scales. The first axis of community composition followed a gradient from high to low cover of

local woody vegetation (Fig. 2.3b). The second axis described a gradient from sites with high

woody vegetation cover at the landscape and context scales, and high context land cover diversity,

to sites in less heterogeneous, pasture-dominated landscapes with low woody vegetation cover (Fig.

2.3b).


43

Figure 2.3 Detrended correspondence analysis (DCA) of bird species composition in: (a) all three land-use

types sampled, and (b) in farmland sites only, with significant environmental variables superimposed on the

ordination plot. Abbreviations for the spatial scales: LC = local scale, CT = context scale, LS = landscape

scale; and for environmental variables: WVC = woody vegetation cover, HET = Heterogeneity, TWI =

Terrain Wetness Index, LCD = Land Cover Diversity, TR = Terrain Ruggedness, PC = Pasture Cover.

Species richness

Species richness increased with increasing local woody vegetation cover, but reached an asymptote

at high woody vegetation cover (Table 2.2; Fig. 2.4). Furthermore, the positive effect of local

woody vegetation cover on species richness was more pronounced in arable fields than in

grasslands (Table 2.2; Fig. 2.4). Richness of species groups with different habitat specialization

showed different and sometimes opposite responses to the environmental variables (Table 2.2).

Richness of farmland birds and forest specialists (in farmland) peaked at intermediate levels of local

woody vegetation cover. In contrast, richness of farmland birds decreased with increasing

landscape woody vegetation cover, whereas the richness of forest specialists increased in response

to this variable. Additionally, richness of forest specialists was positively related to local

heterogeneity. Richness of open-country birds decreased with context ruggedness and woody

vegetation cover (Table 2.2).

Chapter II

44

Table 2.2 Final generalized linear mixed effects models obtained through backward selection (threshold of p

< 0.1) for total species richness and for species richness of groups with different habitat specialization in

arable land and grasslands. The top row for each parameter represents the estimate ± SE.

Total

richness

Forest

specialists

Farmland

birds

Open-country

birds

(Intercept)a 1.64 ± 0.08

p < 0.001

0.42 ± 0.12

p < 0.001

0.76 ± 0.12

p < 0.001

Local Land-use: grassland 0.13 ± 0.09

p = 0.152

0.48 ± 0.14

p < 0.001

Woody vegetation cover 0.60 ± 0.07

p < 0.001

1.13 ± 0.15

p < 0.001

0.84 ± 0.13

p < 0.001

Woody vegetation

cover^2

-0.13 ± 0.05

p = 0.017

-0.43 ± 0.11

p < 0.001

-0.31 ± 0.13

p < 0.001

Land-use Grassland *

woody vegetation cover

-0.29 ± 0.10

p = 0.003

-0.75 ± 0.17

p < 0.001

Heterogeneity

0.24 ± 0.09

p = 0.004

Context Terrain ruggedness 0.16 ± 0.08

p = 0.035

-0.19 ± 0.11

p = 0.089

Woody vegetation

Cover

-0.22 ± 0.10

p = 0.026

Landscape Woody vegetation cover 0.19 ± 0.08

p = 0.020

-0.20 ± 0.06

p = 0.001

Random

effect

Village catchmentb 0.000 0.000 0.000 0.000

Site-levelb 0.021 0.085 0.000 0.000 aResults are given with arable land-use as baseline (i.e. the intercept shown is associated with the regression

line for arable land) bThe rows ‘Village catchment’ and ‘Site-level’ show the variance associated with the random effects

Figure 2.4 Predicted total bird species richness (solid lines) in response to woody vegetation cover in arable

versus grassland sites based on a generalized linear mixed effects model. Open circles (arable sites) and

triangles (grassland sites) represent observed species richness.


45

Discussion

To the best of our knowledge, this is the first regional-scale study on birds in a transforming

traditional farming landscape that is based on a stratified, randomized design with replicated survey

sites spanning the entire agricultural mosaic. Our study shows the high conservation value of

traditional farmland: although alpha diversity of birds was higher in forests than in farmland, the

opposite was the case for beta diversity. That is, forest sites appeared relatively homogenous in bird

community composition, whereas farmland sites showed larger differences between sites. Within

farmland, bird distribution patterns were primarily driven by gradients in woody vegetation cover,

and to a lesser extent by gradients of landscape heterogeneity – rather than by differences in land-

use per se (i.e. grassland versus arable land). In combination, our findings suggest that gradients in

woody vegetation cover (and to a lesser extent in heterogeneity) need to be maintained at multiple

scales to conserve regional bird diversity.

The most important driver of bird community composition in farmland was woody

vegetation cover. As local woody vegetation cover decreased, farmland community composition

changed from birds associated with forest and farmland birds (e.g. the Blue Tit, Parus major, and the

Long-tailed Tit, Aegithalos caudatus), to a community mainly comprising open-country birds (e.g. the

Whinchat, Saxicola rubetra, and the Quail, Coturnix coturnix (see Fig. S2.1)). A similar change in

community composition was observed along the gradient from high context-scale land cover

diversity and woody vegetation cover to less heterogeneous, open pasture-dominated areas (see Fig.

S2.1). Gradients in woody vegetation cover are major drivers of bird community composition

worldwide (Ranganathan et al. 2008; Hanspach et al. 2011), and within Europe, evidence continues

to accumulate that both open farmland and farmland with woody vegetation need to be maintained

and restored to conserve bird diversity in its entirety (Batary et al. 2011b; Fischer et al. 2011).

Species richness followed similar patterns to community composition. Total species

richness was positively related to local woody vegetation cover, although this relationship was less

pronounced in grassland. This difference probably resulted from the greater prominence of several

open-country specialists in grassland (e.g. Saxicola sp.), which avoid woody vegetation (Sanderson et

al. 2013). Bird species richness has been observed to increase with high woody vegetation cover

elsewhere in Europe (e.g. Sanderson et al. 2009), with most research focusing on edges and field

margins (Herzon et al. 2008; Batáry et al. 2010; Wuczynski et al. 2011). The positive response to

woody vegetation typically has been ascribed to a greater diversity of nesting, sheltering and feeding

sites in areas with woody vegetation (Hinsley & Bellamy 2000). The strong effect of local woody

vegetation cover in our study therefore may be related to the small home ranges (<1 to 4 ha, for

most observed birds (Cramp 2000)) of breeding birds, with nesting and foraging site selection likely

determined by fine-scale habitat characteristics. These findings suggest that retaining structures of

woody vegetation at small scales could potentially ameliorate the negative effects of agricultural

intensification on bird richness.

Chapter II

46

Despite the strong effect of local woody vegetation cover on species richness, this effect

was found to level off at high levels of woody vegetation, or become negative in grassland. Land

abandonment, and its subsequent transformation to shrubland and eventually forests, is a major

threat to farmland in Eastern Europe (Kuemmerle et al. 2008) and is expected to expand in the

future (Renwick et al. 2013). This is concerning, because many farmland birds, such as the Red-

backed Shrike (Lanius collurio), need both open areas for foraging and trees or shrubs for breeding,

and thus require landscapes with scattered woody vegetation cover (Brambilla et al. 2010). Thus,

although an increase in woody vegetation might initially enhance bird diversity, a major expansion

of shrubland will likely have negative impacts on farmland bird richness, especially in grasslands.

Context and landscape woody vegetation cover also had significant effects on the richness of birds

with different habitat requirements. Unsurprisingly, open country-specialists were negatively

affected by high woody vegetation cover at the context scale. These species often breed or forage

in open landscapes, and dense cover of woody vegetation could also increase nest predation of

ground nesting birds (Morris & Gilroy 2008). Forest specialists and farmland birds showed

opposite patterns to landscape woody vegetation cover, with farmland birds responding negatively

and forest specialists positively. Thus, besides a general decrease in species richness following local

land abandonment, the abandonment of larger farmland areas would be associated with the

replacement of open-country farmland birds and farmland birds by forest specialists. Such a change

in bird community composition would also encompass the loss of several Species of European

Conservation Concern (Table S2.3). Thus, although extensification through land abandonment may

be a desirable conservation strategy in some intensively managed landscapes (Gregory et al. 2005;

Wretenberg et al. 2007), it could be detrimental to the bird fauna of traditional low-intensity

farming landscapes (Suárez-Seoane et al. 2002; Laiolo et al. 2004; Verhulst et al. 2004; Sirami et al.

2008).

Heterogeneity has been proposed as a major factor underpinning farmland biodiversity

(Benton et al. 2003), and positive effects of land cover diversity and structural landscape features

on bird assemblages have been widely reported (Atkinson et al. 2002; Piha et al. 2007; Sanderson et

al. 2009; Wretenberg et al. 2010; Guerrero et al. 2012). The overall lack of an effect of

heterogeneity in our study (except for the positive response of forest specialists to local

heterogeneity) may therefore be surprising. However, unlike our study, most other studies to date

have not decoupled the effects of woody vegetation cover and land cover heterogeneity. Our

findings suggest that woody vegetation cover may be more important for bird species richness than

other aspects of heterogeneity related to land cover. This possible explanation is consistent with

Sanderson et al. (2009), who found a stronger positive response of bird richness and abundance to

woody edge habitat than to compositional land cover heterogeneity in Poland.

Open-country birds in particular (such as the Skylark, Alauda arvensis) often prefer simpler

landscapes dominated by agriculture and may be expected to respond negatively to heterogeneity

(e.g. Fischer et al. 2011). Nevertheless, within agricultural landscapes, small field sizes, high crop


47

diversity, and a high proportion of field margins provide better conditions for foraging, nesting,

and shelter (Vickery et al. 2001; Guerrero et al. 2012; Josefsson et al. 2013; Kuiper et al. 2013). The

lack of a response to heterogeneity in this group might be related to the coarse resolution of the

CLC map, used to derive context and landscape heterogeneity measures, which only distinguishes

between the main land-use types but not individual arable fields. Alternatively, the reduced

invertebrate prey due to simultaneous application of fertilizers and pesticides in intensified

landscapes may partly explain the loss of open-country farmland birds in the more homogenous

intensified landscapes (Verhulst et al. 2004). In contrast, in our study area pesticide and fertilizer

use is very limited even in the more intensified regions, and hence invertebrates may be rich

throughout the landscape and dampen the effect of heterogeneity on open-farmland specialists.

In combination, our findings suggest that the key mechanism underpinning Transylvania’s

bird community is the availability of gradients in woody vegetation cover and, to a lesser extent,

land cover heterogeneity across multiple different scales. Notably, we captured responses of bird

richness and composition only to those landscape features we could readily quantify, namely woody

vegetation cover and heterogeneity. While these features are important, many other characteristics

of Transylvanian farmland may also be relevant. These include a high variety of crops and crop

rotation within small fields (Guerrero et al. 2012), traditional livestock rotation (Söderström et al.

2001), occurrence of fallow land and semi-natural vegetation (Devictor & Jiguet 2007; Sanderson et

al. 2013), low pesticide and fertilizer use (Kleijn et al. 2009; Geiger et al. 2010), a wide variety of

different field margins (Wuczynski et al. 2011), and land-use that is not optimized for a single

performance or to produce maximum yield (Donald et al. 2001). As observed in other regions

(Doxa et al. 2010; Wretenberg et al. 2010; Sanderson et al. 2013), the loss of low-intensity mixed

farmland from Transylvania very likely would have a negative impact on bird diversity.

Conservation implications

Conservation actions in traditional farming landscapes are urgently needed because farmland bird

populations are declining in Central and Eastern Europe (Reif et al. 2008; Voříšek et al. 2010).

Although national conservation efforts improve population trends, decreasing trends are not being

reversed (Koleček et al. 2014). Research on farmland biodiversity in these traditional farming

landscapes is lagging behind research in Western Europe, which is problematic because the

conservation challenges here may be different from those in more intensified landscapes (Baldi &

Batary 2011; Tryjanowski et al. 2011). Our findings suggest that the key to conserve Transylvania’s

bird diversity is to maintain broad gradients of woody vegetation cover across multiple spatial

scales, and to a lesser extent, maintain small-scale compositional heterogeneity.

In Transylvania, land abandonment is most likely to occur in the grasslands of more

remote, forested, and steep areas. Preventing abandonment in these areas could be achieved by

providing incentives to maintain rotational livestock grazing and to remove shrubs in order to

maintain landscapes with little to intermediate woody vegetation cover. Intensification, on the other

Chapter II

48

hand, is more likely to occur in arable land, especially in the broader, accessible valleys. Here,

conservation strategies should prioritize the retention of woody vegetation cover at intermediate

levels, for example by maintaining various types of field margins (Wuczynski et al. 2011), and

encouraging high land cover diversity through mixed farming. Furthermore, care should be taken in

both grasslands and arable fields that larger tracts of open landscape with low woody vegetation

cover remain available for open-country birds.

In conclusion, there is no simple conservation recipe that can be implemented across

traditional farmland, but rather, farmland bird communities will benefit most from policies targeted

to either mitigate land abandonment or land-use intensification, depending on the specific location.

Because traditional farming has become largely unviable, new strategies need to be identified that

not only maintain farmland biodiversity but also benefit local people, thus preventing an exodus of

the rural population and subsequent cessation of farming. In our study area, potential strategies

could include the uptake of modern agro-ecological and organic farming methods instead of

conventional land-use intensification (Hole et al. 2005; Gabriel et al. 2010), or the development of

incentives for smallholder farmers to engage in (agro-)ecotourism (Young et al. 2010).

Acknowledgements

We thank A. Klein and R. Bigonneau for their tremendous efforts in the field. We are also grateful

to many local people and stakeholders, including the Mihai Eminescu Trust and local government

representatives. We thank P. Griffiths, J. Knorn and T. Kuemmerle for their help with creating the

woody vegetation layer, J. Stein for help with the TWI layer, and D. Abson, P. Fust, and H. von

Wehrden for analytical assistance. We thank Dale Nimmo for constructive suggestions on a

previous draft of this manuscript. The project was funded by a Sofja Kovalevskaja Award by the

Alexander von Humboldt Foundation to JF.


49

Supplementary Material for Chapter II

Table S2.1 The number of replicated combinations of villages according to stratification by landscape terrain

ruggedness and protection status. Terrain ruggedness was classified into three groups based on terciles

covering the entire study area. Sites located within the protected Natura 2000 network are indicated by Sites

of Community Importance (SCI, according to the EU Habitats Directive) and Special Protection Areas (SPA,

according to EU Birds Directive).

Table S2.2 The number of replicated combinations of sites according to stratification by local heterogeneity

and woody vegetation cover in arable (A) and grassland (G) sites. Heterogeneity and woody vegetation cover

were classified into three groups based on terciles covering the entire study area.

Table S2.3 List of recorded bird species, their associated SPEC category, and their habitat specialization.

SPEC 1 represents European species of global conservation concern classified as Critically endangered,

Endangered, Vulnerable, Near Threatened or Data Deficient under the IUCN Red List Criteria at a global

level; SPEC 2 represents species whose global populations are concentrated in Europe and which have an

Unfavorable Conservation Status in Europe; SPEC 3 represents species whose global populations are not

concentrated in Europe but which have an Unfavorable Conservation Status in Europe; and SPEC 4

represents species whose global populations are concentrated in Europe (more than 50% of their global

population or range in Europe) but which have a Favorable Conservation Status in Europe.

Latin English SPEC Habitat

Aegithalos caudatus Long-tailed Tit NON Forest

Certhia familiaris Eurasian Treecreeper NON Forest

Coccothraustes coccothraustes Hawfinch NON Forest

Columba oenas Stock Pigeon 4 Forest

Columba palumbus Wood Pigeon 4 Forest

Cuculus canorus Common Cuckoo NON Forest

Dendrocopos major Great Spotted Woodpecker NON Forest

Dendrocopos medius Middle Spotted Woodpecker 4 Forest

Dendrocopos minor Lesser Spotted Woodpecker NON Forest

Dryocopus martius Black Woodpecker NON Forest

Erithacus rubecula European Robin 4 Forest

Ficedula albicollis Collared Flycatcher 4 Forest

Fringilla coelebs Chaffinch 4 Forest

Garrulus glandarius Eurasian Jay NON Forest

Oriolus oriolus Golden Oriole NON Forest

Parus caeruleus Blue Tit 4 Forest

Parus major Great Tit NON Forest

SCI SPA Non-protected

Terrain Low 1 4 4

ruggedness Medium 4 3 3

High 4 3 4

Heterogeneity

Low Medium High

A G A G A G

Woody vegetation cover Low 11 8 6 8 10 6

Medium 6 7 7 7 6 9

High 6 8 8 7

Chapter II

50

Parus palustris Marsh Tit NON Forest

Phylloscopus collybita Chiffchaff NON Forest

Phylloscopus sibilatrix Wood Warbler 4 Forest

Picus canus Grey-headed Woodpecker 3 Forest

Picus viridis Green Woodpecker 2 Forest

Sitta europea Nuthatch NON Forest

Strix uralensis Ural Owl NON Forest

Sturnus vulgaris Common Starling NON Forest

Sylvia atricapilla Blackcap 4 Forest

Troglodytes troglodytes Eurasian Wren NON Forest

Turdus merula Common Blackbird 4 Forest

Turdus philomelos Song Thrush 4 Forest

Anthus trivialis Tree Pipit NON Farmland

Carduelis cannabina Linnet 4 Farmland

Carduelis carduelis Goldfinch NON Farmland

Carduelis chloris Greenfinch 4 Farmland

Emberiza citrinella Yellowhammer 4 Farmland

Hippolais pallida Eastern Olivaceous Warbler 3 Farmland

Lanius collurio Red-backed Shrike 3 Farmland

Lanius excubitor Great Grey Shrike 3 Farmland

Lanius minor Lesser Grey Shrike 2 Farmland

Locustella fluviatilis River Warbler 4 Farmland

Lullula arborea Woodlark 2 Farmland

Luscinia luscinia Thrush Nightingale 4 Farmland

Passer montanus Tree Sparrow NON Farmland

Phylloscopus trochilus Willow Warbler NON Farmland

Pica pica Magpie NON Farmland

Sylvia borin Garden Warbler 4 Farmland

Sylvia communis Common Whitethroat 4 Farmland

Sylvia curruca Lesser Whitethroat NON Farmland

Sylvia nisoria Barred Warbler 4 Farmland

Turdus viscivorus Mistle Thrush 4 Farmland

Upupa epops Hoopoe NON Farmland

Acrocephalus palustris Marsh warbler 4 Open-country

Alauda arvensis Skylark 3 Open-country

Anthus campestris Tawny Pipit 3 Open-country

Coturnix coturnix Common Quail 3 Open-country

Crex crex Corncrake 1 Open-country

Emberiza calandra Corn Bunting 4 Open-country

Motacilla alba White Wagtail NON Open-country

Motacilla flava Yellow Wagtail NON Open-country

Saxicola rubetra Winchat 4 Open-country

Saxicola torquata Stonechat 3 Open-country

Passer domesticus House Sparrow NON Sinantropic


51

Figure S2.1 Species score of the detrended correspondence analysis (DCA; Figure 2.3b) of bird species

composition in arable and grassland sites.

52

53

Chapter III

54

55

©Ian Lewington

Chapter III

Impact of land cover homogenization on the Corncrake

(Crex crex) in traditional farmland

Ine Dorresteijn, Lucas Teixeira, Henrik von Wehrden, Jacqueline Loos, Jan Hanspach,

John A.R. Stein, Joern Fischer

Landscape Ecology (2015), DOI 10.1007/s10980-015-0203-7

The Corncrake Heard No More

My memory take me back long years to when I was young boy

To evenings in mid-Summer in June and in July

The corncrake called in darkened mead the same notes o'er and o'er

But now in part of Ireland where I lived the corncrake heard no more.

In most of their old breeding grounds the corncrakes now don't breed

The earlier cutting of the grass stripped cover they did need

The silage harvester took its toll their nests and eggs destroyed

And in green meadow near my home the voice of corncrake died

The corncrake's voice no longer heard in meads of Duhallow

And I've not heard their familiar calls for thirty years or so

The earlier cutting of the grass left the birds with nowhere to hide

And the corncrakes have disappeared from my native countryside

On summer evenings long ago some hours after nightfall

In darkened meadows near my home the corncrakes did call

But the migrant rail no longer heard lost to posterity

And that voice I loved when I was young now just a memory

– Francis Duggan

http://www.poemhunter.com/francis-duggan/poems/

Chapter III

56

Corncrakes in traditional farming landscapes

57

Abstract

Context.

The loss of landscape heterogeneity is causing declines of farmland biodiversity around the world.

Traditional farmland regions are often highly heterogeneous and harbor high biodiversity, but are

under threat of land cover homogenization due to changing agricultural practices. One species

potentially affected by landscape homogenization is the Corncrake (Crex crex), which is threatened

in Western Europe but remains widespread in the traditional farmland regions of Eastern Europe.

Objectives.

In this study we present a case study aiming to assess the potential threat of landscape

homogenization for the Corncrake in Romania. We first examined current Corncrake distribution

in relation to woody vegetation cover, landscape heterogeneity, and topography (measured at three

different spatial scales), as well as human disturbance, throughout the existing agricultural mosaic.

Second, we predicted potential future distribution of suitable Corncrake habitat in response to land

cover homogenization by simulating a reduction in land cover diversity.

Results.

Corncrakes were present in grassland and arable fields, and preferentially occupied remote areas

that were wet and flat, and had high land cover diversity at the 100 ha scale. The simulation of land

cover diversity loss revealed that even a moderate reduction of land cover diversity could drastically

reduce the extent of suitable Corncrake habitat.

Conclusions.

Our findings show the high conservation value of traditional heterogeneous farmland for the

Corncrake. Therefore, to protect the Corncrake in traditional farmland, pro-active policy measures

should encourage the continuation of mixed farming practices to maintain a diversity of land

covers.

Chapter III

58

Introduction

Agricultural intensification since the mid-1900s has caused worldwide declines of farmland

biodiversity (Donald et al. 2001; Tilman et al. 2001; Stoate et al. 2009). The process of agricultural

intensification typically involves a combination of changes in management practices to increase

production, such as increased mechanization and use of agrochemicals, loss of crop varieties and

crop rotation, and the decrease of fallow land, non-cropped areas, and natural and semi-natural

vegetation (e.g. Tilman et al. 2002; Tscharntke et al. 2005) . One of the defining consequences of

agricultural intensification is the homogenization of the agricultural mosaic at multiple spatial

scales, for instance through the loss of land cover diversity and structural heterogeneity (Benton et

al. 2003; Tscharntke et al. 2005). Such land cover homogenization has severely reduced the

suitability of farmland for numerous species across multiple taxa through reducing (nesting) habitat,

food availability, and causing habitat fragmentation (Hinsley 2000; Weibull et al. 2003; Concepción

et al. 2008; Guerrero et al. 2012; Tscharntke et al. 2012).

To mitigate the loss of farmland biodiversity, a range of national and international

conservation policies are being implemented. In the European Union, agri-environment schemes

(AES) of the European Common Agricultural Policy (CAP; Pillar 2) are a particularly prominent

strategy (Skogstad and Verdun 2010). Farmers can take up AES voluntarily and receive financial

support to improve habitat quality and promote certain species. However, the effectiveness of AES

for biodiversity conservation has been questioned, and European farmland biodiversity continues

to decline (Kleijn et al. 2006; Kleijn et al. 2011; Pe'er et al. 2014). The apparent ineffectiveness of

AES partly could be explained by the lack of effective tools to assess the outcomes of AES (Kleijn

& Sutherland 2003), or by a poor fit of AES to the ecological requirements of specific farmland

bird species (Princé et al. 2012). Moreover, the effectiveness of AES is likely to be dependent on

context conditions, including levels of land-use intensification (Princé et al. 2012) and landscape

complexity. For example, local-scale AES may be more effective in relatively homogenous

landscapes than in highly heterogeneous landscapes (Tscharntke et al. 2005; Concepción et al. 2008;

Batary et al. 2011a). However, highly heterogeneous landscapes are threatened as a whole

(Tscharntke et al. 2005). To more effectively protect the ecological values of heterogeneous

landscapes, a key challenge therefore is to scale up AES, from local-level measures to landscape-

level measures (Concepción et al. 2008).

The increasing shift in focus from local to larger scale agri-environment measures (e.g.

Dallimer et al. 2010; McKenzie et al. 2013) could thus be of particular interest for heterogeneous

landscapes, such as the traditional farming landscapes in Eastern Europe (Fischer et al. 2012).

Traditional farmland is typically characterized by high heterogeneity in land cover and structural

elements (Plieninger et al. 2006), and subsequently often harbors high levels of biodiversity,

including threatened species (Tscharntke et al. 2005; Kleijn et al. 2009). However, traditional

farming landscapes are under severe pressure of landscape homogenization from changes

associated with agricultural intensification (Lepers et al. 2005). Nevertheless, empirical research to


59

inform conservation management is lacking from the traditional farming regions in Eastern

Europe, and therefore current measures within the CAP are poorly adapted to Eastern European

conditions (Baldi & Batary 2011; Tryjanowski et al. 2011; Sutcliffe et al. 2014). Given this context,

understanding the potential effects of land cover homogenization on biodiversity in traditional

landscapes is an important research priority.

Here, we present a case study highlighting the potential consequence of homogenization of

a traditional farmland mosaic for the Corncrake (Crex crex) in Transylvania, Romania. The

Corncrake is a protected bird species in the European Union that primarily occurs in low-intensity

grasslands and wet meadows with tall vegetation (Green et al. 1997). Agricultural intensification

caused major declines in Corncrake populations in Western Europe (Green et al. 1997), although

recently, some populations have begun to recover (Koffijberg & Schäffer 2006; Koleček et al.

2014). In contrast, large populations persist in former USSR and Eastern European countries.

However, most populations in this region are declining (Koffijberg & Schäffer 2006; Koleček et al.

2014), and current trends in economic development are expected to enhance landscape

homogenization and thus further exacerbate existing declines (Fourcade et al. 2013). To date,

Romania has persisted as a major stronghold of the Corncrake, supporting an estimated 60,000

individuals (Koffijberg & Schäffer 2006).

With this study, we sought to understand the drivers underpinning Corncrake distribution

in traditional Romanian farmland to gauge its sensitivity to possible future land cover

homogenization. Specifically, (1) we assessed the response of Corncrake presence to human

disturbance, and different features of landscape structure at three spatial scales; and (2) based on

this, we simulated the possible effects of land cover homogenization on the distribution of suitable

Corncrake habitat.

Methods

Study area and design

The study area covered 3100 km2 and was located in Southern Transylvania, Romania, at altitudes

between 230 and 1100 m (Fig. 3.1a). Transylvania is one of Europe’s last regions dominated by

traditional, small-scale farming systems. The landscape supports a heterogeneous farmland mosaic

with similar amounts of forest and farmland: 47% forest, 20% pastures, and 25% arable land

(Corine Land Cover Digital map 2006). Arable land is used mainly for low-intensity semi-

subsistence farming, with most fields smaller than two hectares. Most fields are still tilled by horses

and people, and agrochemical use is low by European standards. Considering all farmland, the use

of inorganic fertilizer was only 16 kg/ha on average in 2013 in the four counties in which our study

was carried out (INS 2015), while the average use in Romania is 28 kg/ha and 76 kg/ha for the

entire European Union (www.ec.euopa.eu). Hay meadows provide fodder for livestock and many

are still cut by hand. This, in turn, provides a variety of different sward heights throughout the

spring and summer. The semi-natural pastures are grazed by sheep (dominant livestock), goats and

Chapter III

60

cattle. Semi-natural vegetation is present throughout the landscape as hedgerows, streamside

vegetation, and scattered trees. About half of the study area (1500 km2) was located within the EU

protected Natura 2000 network, including both Sites of Community Importance (SCI, Habitats

Directive), and Special Protection Areas (SPA, Birds Directive; Fig. 3.1a).

Figure 3.1 A) The inset shows the location of Romania in Europe (red) and the location of the study area in

Romania (black). The close-up of the study area shows the 42 focal villages, location of the survey sites, and

the area protected in the Natura 2000 network. The altitude is given for the area occupied by farmland

(arable land and pastures), and represents the area for predicted Corncrake habitat suitability in Figure 3.3. B)

The location of Corncrake presences in relation to land cover diversity at the context scale (100 ha).

To select our survey sites, we divided the study area into ‘village catchments’, defined as

villages including their surrounding land. Such village catchments have been identified as both

ecologically and socially meaningful landscape units (Angelstam et al. 2003). We delineated the area

belonging to a given village catchment using a cost-distance algorithm that allocated each pixel to

the village with the lowest travel cost (defined by slope) to this pixel in ArcGIS 10.1. The study area

contained 172 villages. We randomly selected a subset of 42 villages (Fig. 3.1a), stratified to cover

the full gradients of terrain ruggedness (low, medium, high) and protection level (Natura 2000, no

protection). Terrain ruggedness was divided into three categories based on terciles of ruggedness

values across the entire study area. The standard deviation ranged from 32.5 m to 47.8 m, with a

mean altitude of 491 m for the low category; from 47.8 m to 57.7 m with a mean altitude of 500 m

for the medium category; and from 57.7 m to 97.2 m, with a mean altitude of 518 m for the high

category. Thus, seven villages were selected for each possible cross-combination of ruggedness

level and protection status. Within each village, survey sites were placed along all drivable roads

with a distance of 500 m between sites (Moga et al. 2010; Fig. 3.1a), resulting in 331 survey sites in

total. The number of sites varied between 2 and 18 per village catchment.


61

Corncrake surveys

Corncrake surveys were performed under suitable weather conditions in the breeding season of

2013, from May 15 to June 30, from 23:00 to 3:00 hours, which is the most active calling period

(Moga et al. 2010). Each site was surveyed twice, using five minute point counts from the road

based on auditory observations (Bibby et al. 2000). The Corncrake typically calls loudly and

frequently, and is often heard over several hundred meters. Five minutes appeared to be an

appropriate time span, with 82% of all males detected within one minute and 98% within four

minutes. The location of each singing (male) Corncrake was calculated by triangulation, based on

two to three GPS points and their respective compass bearing towards the singing bird.

Triangulation was implemented in ArcGIS using the tool ‘Bearing distance to line’; Corncrake

locations were defined as the crossing of two lines in case of two GPS points or as the center of

the polygon in case of three GPS points.

To broadly characterize the landscape surrounding Corncrake locations we identified the

land cover of all Corncrake locations from the road within the following categories: arable land,

fallow land, hay meadow, pasture, and marginal elements (e.g. reed patches and field margins). No

further habitat measurements were taken because detailed microhabitat requirements of the

Corncrake in this region were previously described by Moga et al. (2010).

Variables used to model Corncrake presence

We modelled Corncrake presence/absence around each site in relation to two indices of human

disturbance as well as to environmental variables that may explain Corncrake occurrence. The

indices of human disturbance were human population density and distance to village. Environmental

variables were measured at three spatial scales. The local scale (circular area of 10 ha, radius of 178

m) covered the approximate home range of the Corncrake (Grabovsky 1993; Van Weperen 2009),

which rarely moves more than 250 m between calling sites (Peake & McGregor 2001). The context

scale (circular area of 100 ha, radius of 564 m) and village catchment scale (mean ± SD: 18.2 km2 ± 10.7

km2) represented scales beyond the typical daily movements of male Corncrakes (Sklíba & Fuchs

2004).

At each scale, we quantified environmental variables expressing woody vegetation cover,

land cover heterogeneity, and topography (Table 3.1). Both woody vegetation cover and land cover

heterogeneity are likely to be affected by possible land cover homogenization (Benton et al. 2003;

Tscharntke et al. 2005). In contrast, topography controls key biophysical gradients within the

landscape, but is not subject to medium-term changes.

At the local scale we included the variables: (i) proportion of woody vegetation cover, (ii)

spectral variance of monochromatic SPOT imagery (©CNES 2007, Distribution Spot Image SA) as

an index of local structural heterogeneity, and (iii) Topographic Wetness Index (TWI) to indicate

topographic position and potential soil wetness. At the context scale we included: (i) Simpson’s

diversity index of land cover (assessed with FRAGSTATS 4.1, McGarigal et al. 2012), and (ii)

Chapter III

62

terrain ruggedness. We excluded woody vegetation cover at this scale because it was positively

correlated with terrain ruggedness (ρ = 0.63). We also excluded edge density (see below) because it

was positively correlated with the Simpson’s diversity index of land cover (ρ = 0.75). Finally, at the

village catchment scale we included: (i) Simpson’s diversity index of land cover, (ii) land cover edge

density, (iii) proportion of arable land cover, and (iv) terrain ruggedness. Together, these

(statistically independent) variables characterized major differences in land cover composition and

configuration between village catchments. We did not include woody vegetation at this scale

because it is dominated by forest cover which is unsuitable for Corncrakes. For a detailed

description of human disturbance and environmental variables and their calculation, see Table 3.1.

Statistical analysis

We pooled Corncrake presence data over the two repeated surveys. Because some individual

Corncrake locations were in very close proximity to one another, we did not attempt to distinguish

between individuals between the two repeats. At sites where no Corncrake was detected (n = 171),

one absence point was randomly allocated in farmland within a 1000 m circular buffer, which

equals the maximum distance a Corncrake call can be heard (Moga et al. 2010).

All analyses were implemented in the R-environment using the packages ‘MASS’, ‘raster’,

‘rgdal’, ‘maptools’ and ‘spatstat’ (R Development Core Team 2013). We modeled Corncrake

presence/absence using a generalized linear mixed effects model (function glmmPQL from the

‘MASS’ package) with a binomial error distribution. The models included all human disturbance

and environmental variables as explanatory variables (fixed effects), village as a random effect to

account for nested experimental units, and an autocorrelation structure of order 1, implementing

geographic coordinates to account for spatial autocorrelation (Dormann et al. 2007). The final

minimum adequate model was obtained through stepwise backward selection (p ≤ 0.1), because

AIC is not implemented in glmmPQL. The model was internally validated using AUC based on a

model calibrated on a randomly selected training dataset (70% of the data) and predictions to the

remaining validation dataset (30% of the data). Validation was repeated 100 times and AUC values

were averaged. We confirmed that variables were not correlated (ρ < 0.6), log-transformed local

woody vegetation cover and population density, square-root transformed distance to village, and

standardized all variables to zero mean and unit variance, prior to modelling.


63

Table 3.1 Description and calculation of human disturbance variables and the environmental variables used

to model Corncrake distribution. Environmental variables were calculated at three spatial scales. The data

source for the environmental variables and units for variables with a unit are given.

Based on the model outlined above, we simulated how the distribution of suitable

Corncrake habitat would change under hypothetical scenarios of land cover homogenization. This

was implemented by a stepwise reduction of context land cover diversity and subsequently

predicting suitable Corncrake habitat for each step. Notably, we only altered context land cover

diversity, because of the modelled variables it is the variable most likely to change under land cover

Chapter III

64

homogenization. Other variables that influenced Corncrake occurrence are either not influenced by

land cover change (e.g. ruggedness), or would change in unpredictable ways (e.g. human population

density). We reduced land cover diversity incrementally by values of 0.01 for a given pixel in a given

step (original dataset: min = 0, max = 0.84). Pixels with a value of zero were not reduced further.

After each homogenization step, we predicted the probability of suitable Corncrake habitat in

farmland (grassland and arable land) of the whole study area using the fixed effects estimates from

the above described Corncrake model. Finally, the predicted probabilities of occurrence were

converted into presence/absence data using a probability threshold of 0.5 to define suitable

Corncrake habitat.

Based on the results of this simulation, we calculated percent loss in land cover diversity as

the proportion of land cover diversity in a given step relative to the current situation and

standardized to 100%. Similarly, we calculated the remaining proportion of suitable Corncrake

habitat based on the area of predicted suitable Corncrake habitat in a given simulation step relative

to the total area of suitable Corncrake habitat in the region. To visualize the effects of reducing

land cover diversity, we plotted percent loss of land cover diversity against the percent of available

suitable Corncrake habitat. Additionally, land cover diversity and Corncrake suitable habitat were

mapped, both for current conditions and for habitat losses of 33 % and 66 %, corresponding to the

loss of a third or two thirds of the available suitable habitat, respectively.

Results

We found 114 Corncrake presences pooled over the two repeats, of which 71 were located in the

Natura 2000 area (Fig. 3.1b). Corncrakes occurred most frequently in arable land (n = 36; 25 in

Alfalfa fields, Medicago sativa subsp. sativa), marginal elements (n = 31; mostly reed (Phragmites

australis)), and hay meadows (n = 25), followed by fallow land (n = 13) and pastures (n = 9).

Corncrake presence was predicted by human disturbance variables and by environmental

variables at the local and context scales (Table 3.2). The Corncrake was more likely to occupy

remote locations at large distances from villages, and village catchments with a low human

population density. Moreover, the species was more likely to occur in areas with low terrain

ruggedness and high land cover diversity at the context scale, and a high topographic wetness index

at the local scale (Table 3.2; Fig. 3.1b). Model validation indicated a good overall model

performance (AUC mean ± SD: 0.84 ± 0.04).


65

Table 3.2 Model coefficient estimates ± Standard Error (SE) of the final generalized linear mixed effects

model obtained through backward selection (threshold of p < 0.1). ‘--' indicates variables not contained in the

final model.

Scale Variable Estimate ± SE p

(Intercept) -0.57 ± 0.37 0.13

Human disturbance Population density -0.77 ± 0.40 0.06

Distance to village 0.35 ± 0.15 0.02

Local Woody vegetation cover -- --

Heterogeneity -- --

Topographic wetness index 0.31 ± 0.14 0.03

Context Land cover diversity 0.49 ± 0.18 <0.01

Terrain ruggedness -0.95 ± 0.20 <0.01

Village catchment Land cover diversity -- --

Edge density -- --

Arable land cover -- --

Terrain ruggedness -- --

The simulation of Corncrake distribution in response to landscape homogenization

showed that the extent of predicted suitable Corncrake habitat decreased rapidly with a simulated

loss of land cover diversity at the context scale. Notably, a substantial contraction of suitable

Corncrake habitat was already apparent at relatively small reductions in land cover diversity (Fig.

3.2, 3.3). For example, an 11% overall decrease in current land cover diversity (based on the

complete range of the variable standardized to 100%) resulted in a 33% loss of suitable Corncrake

habitat (Fig. 3.3; middle row); and a 35% decrease in current land cover diversity resulted in a 66%

loss of Corncrake habitat (Fig. 3.3; bottom row).

Figure 3.2 Predicted suitable Corncrake habitat as a function of land cover diversity loss within the context

scale (100 ha). Land cover diversity loss was simulated through a step-wise reduction of current land cover

diversity. Suitable Corncrake habitat was calculated as the remaining proportion of Corncrake habitat based

on the area of predicted Corncrake habitat in a given simulation step relative to the total area of Corncrake

habitat in the region.

Chapter III

66

Figure 3.3 Predicted distribution of suitable Corncrake habitat under current and reduced land cover

diversity in a 100 ha surrounding (resolution 27 m x 27 m). The left column shows changes in land cover

diversity and the right column shows the predicted concurrent change in the distribution of suitable

Corncrake habitat. The upper row represents the current situation. Thresholds for simulation of land cover

diversity were selected to illustrate a habitat loss of 33% and 66% of the current habitat. These figures were

chosen to exemplify the continuous pattern from Figure 3.2 and were purely based to show a third and two

thirds of habitat lost. The middle row represents an 11% reduction in land cover diversity which reduces

Corncrake habitat by 33%. The lower row represents a 35% reduction in land cover diversity which reduces

Corncrake habitat by 66%. Land cover diversity varies from low (grey) to high (black) on a standardized scale

between 0 and 1. The boxplot below each figure represents the distribution of land cover diversity for a given

step. Suitable Corncrake habitat (black) was assumed by a predicted probability of Corncrake occurrence of

0.5 or higher.


67

Discussion

Our study adds to a growing body of knowledge on the effects of heterogeneity on farmland

biodiversity, suggesting that the loss of heterogeneity, typically associated with land-use

intensification, can negatively impact biodiversity (e.g. Stein et al. 2014). Furthermore, substantial

losses of Corncrake habitat are likely to result from relatively small losses in land cover diversity,

indicating a non-linear relationship between habitat loss and loss in land cover diversity. These

findings have important implications for the management of heterogeneous farming landscapes. In

Europe, the majority of studies on the effects of landscape homogenization on farmland

biodiversity have been performed in countries with more intensively used farmland (Devictor &

Jiguet 2007; Pickett & Siriwardena 2011), while more heterogeneous, low-intensity regions remain

poorly understood (Baldi & Batary 2011; Tryjanowski et al. 2011). Our findings suggest that land

cover homogenization in a traditional farming landscape could drastically reduce suitable habitat

for farmland species such as the Corncrake. This underlines the need for conservation strategies

that consider land cover diversity across entire landscape mosaics, and that are specifically geared

towards traditional farmland supporting large populations of rare or threatened species.

Agricultural landscapes are increasingly recognized for their value for biodiversity

conservation (Mendenhall et al. 2014). This is especially important for European landscapes, which

reflect centuries of interactions between people and the natural environment (Bignal & McCracken

2000). Long-established traditional agriculture has created many of the habitats and niches for

species valued for biodiversity today (Bignal & McCracken 2000). Nevertheless, traditional

agricultural landscapes have declined dramatically over the last decades (Foley et al. 2005). The few

remaining examples are under considerable threat from changes in agricultural management and

their continued existence hinges on both national and international conservation policies (Henle et

al. 2008).

Our study shows that land cover heterogeneity is a key feature to be addressed in

conservation measures targeting traditional farmland. The positive effects of heterogeneity on

biodiversity, including farmland birds, have been well-documented (e.g. Weibull et al. 2000; Benton

et al. 2003; Diacon-Bolli et al. 2012; Stein et al. 2014). In particular, land cover diversity appears to

positively affect bird richness and abundance in farmland (Sanderson et al. 2009; Pickett &

Siriwardena 2011), suggesting that different land-use covers provide complementary resources,

such as feeding, nesting, and sheltering sites. However, responses of farmland birds to

heterogeneity are species-specific; especially ground nesting birds and open farmland specialists

often prefer relatively homogenous landscapes (Filippi-Codaccioni et al. 2010; Fischer et al. 2011;

Pickett & Siriwardena 2011; Guerrero et al. 2012). For such specialist species, management

practices to increase heterogeneity may therefore reduce their populations, especially in low-

intensity farmland (Batary et al. 2011b).

Also the Corncrake is a ground-nesting species in open farmland, and similar to other

grassland species, it has been observed to occur in extensive areas of relatively homogenous,

Chapter III

68

uncultivated meadows (Keišs 2005; Berg & Hiron 2012; Quinn et al. 2012). Nevertheless, a

traditional farmland mosaic such as the one we studied evidently can provide important additional

habitat for the Corncrake. Land cover homogenization at the context scale as consequence of

intensification would have a substantial negative impact on the Corncrake in currently

heterogeneous landscapes. To protect the Corncrake in traditional farmland, it is therefore

important to ensure the persistence of mixed farming practices.

While the use of low-intensity grasslands and hay meadows by the Corncrake is well

documented (Berg & Gustafson 2007; Budka & Osiejuk 2013), less is known about the

complementary values of other land-use types to the species (but see e.g. Berg & Hiron 2012). For

example, in a diverse land cover mosaic, uncropped arable land, and the high density of edges

between different land covers (e.g. field margins or ditches) provide safe breeding or sheltering sites

for the Corncrake (Green et al. 1997; Corbett & Hudson 2010; Budka & Osiejuk 2013). Such

structures are especially important during the early breeding season when the height of

surrounding spring-vegetation is short (Budka & Osiejuk 2013). In addition, field margins are often

rich in invertebrates (Josefsson et al. 2013), and thus a higher abundance of field margins may

increase food availability for the Corncrake.

Some studies have found that the Corncrake tends to avoid annually cropped arable fields

(Keišs 2005; Berg & Hiron 2012; Budka & Osiejuk 2013). However, we found a considerable

number of Corncrakes in arable fields, and in particular in Alfalfa fields. Corncrakes and other

grassland species have also been observed in grass or clover fields used for silage and hay in other

regions (Corbett & Hudson 2010; Berg & Hiron 2012; Quinn et al. 2012). Although this suggests

that arable fields sown with grass or leguminous plants can provide important additional habitat for

Corncrakes, care should be taken for appropriate management because hayed crops can create

population sinks (Perlut et al. 2006). In contrast, pastures appeared to have little habitat value for

the Corncrake, and similarly to other regions, the Corncrake was more likely to be present in hay

meadows than pastures (Wettstein et al. 2001; Keišs 2005). The Corncrake may avoid grazed

pastures for several reasons. For example, they are typically located on better drained, steeper

slopes, whereas the Corncrake preferentially uses moist flat areas. In addition, the reduced

vegetation height in pastures caused by grazing may reduce habitat suitability (Berg & Gustafson

2007).

Loss of hay meadows and the intensification of grassland management are currently

viewed as the main threats to Corncrake populations (Koffijberg & Schäffer 2006). Therefore,

current conservation strategies (including AES schemes) for the Corncrake relate primarily to the

maintenance of hay meadows and low-intensity grasslands, with emphasis on specified mowing

regimes (Koffijberg & Schäffer 2006; Berg & Gustafson 2007) and leaving unmown refuge strips

(Tyler et al. 1998). These methods have been successful (Koffijberg & Schäffer 2006), and have

also provided wider biodiversity benefits in countries with relatively intensive agriculture

(Wilkinson et al. 2012). However, our results suggest that the implementation of grassland-focused


69

strategies alone may not be sufficient for Corncrake conservation in Transylvania’s traditional

farmland. Moreover, at present, farmers in Transylvania mow their land in a way that creates a

heterogeneous vegetation structure, including numerous unmown patches until late in the season.

Delayed mowing regimes could potentially be counter-productive, because they may result in the

synchronization of management and thus homogenization of vegetation height (Dahlström et al.

2013), which has negatively affected Corncrakes elsewhere (Brambilla & Pedrini 2013). Other

specific CAP measures in Romania, such as sheep premium payments, also may pose a threat to the

Corncrake. This is because such incentives may inadvertently encourage the conversion of hay

meadows to pastures (Demeter & Kelemen 2012). These examples highlight the need to carefully

adjust agri-environment schemes and other CAP measures to traditional farmland. Moreover,

schemes particularly related to the Corncrake should primarily target core Corncrake habitat, such

as remote, flat and wet areas, which are especially at risk of conversion to homogenous, intensified

farmland.

Our findings thus underline that agri-environment management needs to be adapted to the

context of the landscape (Batary et al. 2011a), and schemes designed for high-intensity landscapes

cannot be readily transferred to traditional farmland (Sutcliffe et al. 2014). Given the affinity of the

Corncrake with the heterogeneous farmland mosaic, there could be major benefits in agri-

environment schemes specifically aiming at maintaining the heterogeneous character of low-

intensity landscapes (Kleijn et al. 2011). Without pro-active policy measures to maintain land cover

heterogeneity in traditional farmland, homogenization of the landscape will have negative effects

on biodiversity in general, and on already threatened species such as the Corncrake in particular.

The successful implementation of measures that maintain land cover heterogeneity across

entire landscapes will hinge on finding ways to coordinate actions across multiple farmers, rather

than relying on local measures alone (McKenzie et al. 2013). Moreover, high heterogeneity in

traditional farmland is tightly linked to the multi-functionality of semi-subsistence farm holdings of

relatively small size (Tryjanowski et al. 2011). Thus, landscape-scale governance may be challenging

because it depends on the involvement of many individual farmers (McKenzie et al. 2013). In

traditional farmland, AESs could be used to support collaborations of specific target groups (Pe'er

et al. 2014), such as small-scale farm holders which create the heterogeneous mosaic but are also

particularly susceptible to selling land to large-scale farmers.

Conclusions

Our study showed that heterogeneous farmland has high complementary conservation value for

the Corncrake, in addition to their core habitat of large areas of wet meadows. Furthermore, our

simulations highlighted that in heterogeneous low-intensity farmland, conservation measures

implemented on small scales by single farmers should be complemented with landscape-scale

measures to maintain land cover heterogeneity. Although we highlighted the importance of

landscape-scale conservation for a single protected species, a similar pattern appears to be emerging

Chapter III

70

more generally (Merckx et al. 2009; Dallimer et al. 2010). More research is needed on the

effectiveness of landscape-scale conservation in farmland; however, research should also focus on

how conservation measures targeting whole landscapes rather than single farms can be

implemented. This is especially important in traditional farming landscapes, many of which are

complex but rapidly changing social-ecological systems that are threatened by land cover

homogenization.

Acknowledgements

We thank A. Klein and J. Tinnemans for their tremendous effort in the field. We thank P. Griffiths,

J. Knorn and T. Kümmerle for their help with creating the woody vegetation layer, and D. Abson

and P. Fust for analytical assistance. We thank two anonymous reviewers for constructive

suggestions on a previous draft of this manuscript. The project was funded by a Sofja Kovalevskaja

Award by the Alexander von Humboldt Foundation to JF.

71

Chapter IV

72

73

Chapter IV

Human-carnivore coexistence in a traditional rural landscape

Ine Dorresteijn, Jan Hanspach, Attila Kecskés, Hana Latková, Zsófia Mezey, Szilárd Sugár,

Henrik von Wehrden, Joern Fischer

Landscape Ecology (2014), 29:1145-1155

‘There is no conservation without conservation.’ – Luigi Boitani

Chapter IV

74

Human-carnivore coexistence in traditional farming landscapes

75

Abstract

Facilitating human-carnivore coexistence is a major conservation concern in human-dominated

landscapes worldwide. Useful insights could be gained by studying and understanding the dynamics

of human carnivore-coexistence in landscapes in which carnivores and humans have coexisted for a

long time. We used a two-pronged approach combining ecological and social data to study

coexistence of the brown bear (Ursus arctos) and humans in Transylvania, Romania. First, we

surveyed 554 km of walking transects to estimate activity via a bear sign index, namely the

proportion of anthills disturbed by bears, and used spatially explicit predictive models to test which

biophysical and anthropogenic variables influenced bear activity. Second, we interviewed 86

shepherds and 359 villagers and community representatives to assess conflicts with bears and

attitudes of shepherds towards bears. Our interdisciplinary study showed that bears and humans

coexisted relatively peacefully despite occasional conflicts. Coexistence appeared to be facilitated

by: (1) the availability of large forest blocks that are connected to the source population of bears in

the Carpathian Mountains; (2) the use of traditional livestock management to minimize damage

from bears; and (3) some tolerance among shepherds to occasional conflict with bears. In contrast,

bear activity was unrelated to human settlements, and compensation for livestock losses did not

influence people’s attitudes toward bears. Our study shows that coexistence of humans and

carnivores is possible, even without direct economic incentives. A key challenge for settings with a

discontinuous history of human-carnivore coexistence is to reinstate both practices and attitudes

that facilitate coexistence.

Chapter IV

76

Introduction

Facilitating coexistence between humans and carnivores is a conservation challenge worldwide

(Woodroffe et al. 2005; Treves et al. 2006; Dickman et al. 2011), because carnivores have often

been extirpated locally due to conflicts with humans (Breitenmoser 1998; Woodroffe 2000).

Predators are important because they exert top-down control on ecosystem processes (Estes et al.

2011; Ripple et al. 2014), and provide emotional, recreational, and cultural benefits to society

(Kellert et al. 1996). Due to increasing conservation efforts (Ray et al. 2005), some carnivore

populations are growing again (Linnell et al. 2001; Enserink & Vogel 2006), but many continue to

decline (Ripple et al. 2014). The two most frequently advocated strategies to counteract carnivore

declines are: (i) to separate carnivores from settlements by establishing protected refuge areas

(Karanth et al. 2010; Packer et al. 2013); and (ii) to promote human-carnivore coexistence in

human-dominated landscapes through conflict mitigation programs (Woodroffe et al. 2005;

Dickman et al. 2011).

Several studies have shown that humans and carnivores can coexist (Carter et al. 2012b;

Schuette et al. 2013). However, coexistence is often hampered by human-carnivore conflicts, which

can harm rural households especially (e.g. Holmern et al. 2007). Because many carnivores live in

human-dominated landscapes, a key to their successful conservation is to better understand the

dynamics of human-carnivore coexistence. To this end, one useful approach could be to learn from

landscapes in which carnivores and humans have coexisted for a long time.

In Eastern Europe, large carnivores and humans have co-inhabited multiple-use

landscapes for centuries. Romania sustains large, stable populations of the brown bear (Ursus arctos),

wolf (Canis lupus) and lynx (Lynx lynx) (Salvatori et al. 2002), with the brown bear population being

particularly large (estimated at 6000 individuals by the IUCN). Most of Romania’s bears live in the

Carpathian Mountains, but many also occur in the Transylvanian foothills of the Carpathian

Mountains, which harbor hundreds of villages characterized by traditional semi-subsistence

agriculture. This situation is exceptional because the majority of bear populations elsewhere in

Europe are confined to remote mountainous areas – and where bears do range into human-

dominated landscapes, they often damage livestock, orchards and beehives (Zedrosser et al. 2001).

Thus, Transylvania offers an interesting model system that may help to facilitate greater

understanding of the dynamics underpinning successful coexistence of humans and carnivores.

Gathering reliable data on carnivore distribution is notoriously difficult due to their

elusive nature. Especially in Romania, long-term data is scarce, and the reliability of official data

collected by hunting organizations may be questionable (Salvatori et al. 2002). Therefore, we used a

new, sign-based metric, namely the proportion of anthills destroyed by bears relative to the total

number of anthills in an area, being fully aware of the limitations of sign-based indicators (Barea-

Azcon et al. 2007; Long et al. 2007). Although the proportion of destroyed anthills may be a less

accurate sign than footprints or faeces, it offers the opportunity to survey large areas in a relatively


77

short time, which we believe more than compensates the risks of potentially higher methodological

uncertainty.

To understand long-term coexistence in multiple-use landscapes, both ecological and social

variables are important (Treves & Karanth 2003; Treves et al. 2006; Carter et al. 2012a). Although

the need for interdisciplinary work on human-wildlife coexistence has repeatedly been

acknowledged (Redpath et al. 2012), few studies have combined ecological studies on habitat

preferences of carnivores with social data on human-carnivore conflicts and tolerance levels

towards carnivores (Glikman and Frank, 2011; but see Schuette et al. 2013). The overarching goal

of our study was to assess how humans and bears coexist in Southern Transylvania. Our study had

two specific objectives. First, we sought to understand spatial patterns of bear activity in response

to anthropogenic variables, biophysical variables, as well as local connectivity between forest

patches and regional connectivity to the Carpathian Mountains. Second, we examined the nature of

human-bear conflicts in the region and related it to the spatial distribution of bear activity.

Methods


Our 7441 km2 study area was located within a 50 km radius around the town of Sighisoara, in the

foothills of the Carpathian Mountains in Southern Transylvania (Fig. 4.1a). Recently, one of the

largest sets of non-mountainous EU protected Natura 2000 sites was established in this area (Fig.

4.1b). We selected transects inside and outside this protected area to provide data for the

foundations of the brown bear management plan, and to provide reference data so the

management plan can later be evaluated. A detailed description of the area is provided in

Supplementary Material Text S4.1.

To quantify bear activity, we used the proportion of anthills destroyed by bears relative to

the total number of anthills in each pasture transect. Destroyed anthills and other signs are rare in

forests and therefore we walked transects on pasture adjacent to forest. Anthills are available on

pastures across the entire landscape which allowed us to cover a large area in a standardized way.

Ants form an important food source of protein for the brown bear (Dahle et al. 1998; Swenson et

al. 1999). Ant larvae are particularly sought after by bears in spring and summer, and the incidence

of destroyed anthills has been used in previous sign surveys for bear presence (Munro et al. 2006;

Ciarniello et al. 2007). Anthills destroyed by bears are readily distinguishable from those destroyed

by other animals (e.g. cattle) and humans because they have a characteristic ‘crater’ dug out of the

top.

Our choice of transects was guided by three design considerations. First, we sought to

obtain a broad overview of bear activity in the study area. We surveyed bear activity in 30 focal

villages and their surrounding land, because villages have been identified as both ecologically and

socially meaningful ‘landscape’ units (Angelstam et al. 2003). We delineated the area belonging to a

Chapter IV

78

given village using a cost-distance algorithm that allocated each pixel to the village with the lowest

travel cost to this pixel (slope-penalized distance, implemented in ArcGIS, mean area ± SD: 20.5 ±

11.2 km2). We randomly selected 30 villages from 448 villages in the study area, stratified to cover

the full gradient in terrain ruggedness and to include Natura 2000 areas as well as unprotected areas

(Fig. 4.1b). With few exceptions arising from logistical obstacles we surveyed four transects around

each village (n = 113; described in detail below).

Second, we covered the focal conservation area (2675 km2) in detail to inform the new

management plan. We divided the Natura 2000 area into grid cells of 2 km x 2 km (n = 759). We

surveyed one transect in each grid cell with more than 15% forest cover (n = 417), and we surveyed

one transect per grid cell for 100 additional, randomly chosen cells with less than 15% forest cover.

For these two considerations, all transects were chosen to cover the full gradient of available

distances to the nearest village and amount of surrounding forest cover within a radius of 1500 m.

Third, to assess the connectivity from our study area to the assumed source population of brown

bears in the Carpathian Mountains, we compared alternative cost-distance metrics to the

mountains, so that the best cost-distance metric could be used in later analyses (see the section on

variables used to model bear activity below). For this, we specifically surveyed 59 transects within

15-20 km from the mountains (Fig. 4.1b). These transects were chosen to cover a gradient from

likely low cost to high cost by varying the sets of land cover types available between a given

transect and the mountains.

All transects were walked once at the forest-pasture interface at approximately 10 m

distance from the forest edge. Each transect was 800 m long and 6 m wide. All destroyed and

undestroyed anthills within a transect were counted. We included both fresh (this year) and old

(previous years but clearly visible) destroyed anthills. In total, we walked 692 transects (554 km)

between April 28 and August 10, 2012. Three transects were excluded from analysis because they

had no anthills. The mean number of anthills per transect was 139. All transects for the cost-

distance analysis were surveyed in May and June, 2012.


79

Figure 4.1 Study area, study design and predicted bear activity. (A) Study area in southern Transylvania,

Romania, and location of the Carpathian Mountains used for the cost-distance analysis. (B) Close-up of the

study area and study design and location of the transects used for the bear activity model (n = 630) and cost-

distance analysis to the Carpathian Mountains (n = 59). The Natura 2000 area was surveyed for bear activity

in detail. The study area was surveyed at a large scale for bear activity and human-bear conflict in 30 focal

villages and their surrounding land. (C) Predicted bear activity in the study area, based on generalized linear

modelling. Bear activity is indicated as the predicted proportion of destroyed relative to total number of

anthills.

Chapter IV

80

Variables used to model bear activity

We used four sets of variables to explain patterns in bear activity. First, anthropogenic disturbance

was indicated by distance to settlements. Second, the biophysical environment within a radius of

1000 m from the midpoint of each transect was indicated by (i) terrain ruggedness calculated as the

standard deviation of the altitude; (ii) percent pasture cover; and (iii) the shape and size of forest

patches indicated by the forest edge to forest interior ratio. We did not include forest cover because

it was correlated with forest edge to interior ratio (ρ = - 0.71). Third, local connectivity between

forest patches within the study area was indicated by betweenness centrality. Betweenness centrality

is an index of how well connected a given forest patch is within the network of forest patches

regardless of land-use between the different forest patches. Fourth, connectivity to the Carpathian

Mountains (the presumed source population) was indicated by the cost-distance of each transect to

the mountains. Cost-distance was based on a matrix with a resolution of 10 m x 10 m and

combined distance to the mountains with weights between 1-10, assigned by experts to the

different land-use types. Detailed descriptions of these variables are provided in Supplementary

Material Text S4.1.

We accounted for possible effects of the total number of anthills per transect and survey

date (bears might be more attracted to pastures with more anthills; and the number of destroyed

anthills by definition accumulates with time) by including them as variables in the model. Because

the effects of number of anthills and survey date were not necessarily linear, we included their

quadratic terms (e.g. the number of destroyed anthills might level off before the end of the field

season if bears shift to other food sources such as fruit). Protection level was not included in the

models, because the recent establishment of the Natura 2000 area has not yet resulted in different

natural resource use that could affect bear activity.

Statistical modelling

All analyses were performed with the proportion of destroyed vs. total number of anthills as the

response variable within generalized linear models (GLM) with a quasi-binomial error structure to

account for overdispersion. This model specification took into account that the precision of the

response variable increased with an increasing total number of anthills found along a given transect

(e.g. one destroyed anthill out of two total anthills is less precise information than 50 destroyed out

of 100). The final GLM was obtained by using model averaging based on a 10-fold cross-validation

(Fielding & Bell 1997). Validation was done by relating the predicted activity (based on a model

using nine tenths of the data) with the activity observed in the remaining tenth via a binomial GLM

(see Table S4.1 for the estimates and explained deviance for each of the validation steps). We

calculated the amount of deviance explained as a measure of predictive performance of each of the

ten models. We averaged the parameter estimates over the 10 models, weighting the estimates from

each model by its predictive performance. The averaged estimates were used to predict bear activity

in forest edges bordering pastures across the entire study area, at a resolution of 800 m, equalling


81

the transect length of the field surveys. All statistical analysis were implemented in the ‘R’

environment (R Development Core Team 2013). Further details on statistical modelling and

predictive mapping are provided in Supplementary Material Text S4.1.

Human-bear conflicts

Human-bear conflicts were assessed using questionnaires in the same 30 villages targeted for bear

transects (Fig. 4.1b). We used a detailed questionnaire for shepherds that addressed attacks on

livestock, attitudes towards bears, and the current compensation schedule; and a shorter one for

villagers that addressed the types of damage caused by bears. To assess attitudes among different

stakeholder groups relevant to coexistence dynamics, we used the same short questionnaire for

mayors, hunters and local councillors. These groups are in charge of compensation payments after

carnivore damage, and we therefore included several questions regarding compensation

(questionnaires are available as Supplementary Material Text S4.2-4.4). We aimed to interview

approximately three shepherds (on average a village had between one and four shepherds), ten

villagers, one mayor and one hunter or councillor per village, but not everyone was available in all

villages. Ultimately, we obtained questionnaires from 86 shepherds (73 sheep-herders, 13 cow-

herders), 302 villagers, 22 mayors or vice-mayors, 20 local councillors and 15 hunters. (For ethical

considerations see Supplementary Material Text S4.1.)

We expected a significant relationship between bear activity and perceived number of

conflicts (e.g. higher bear activity could increase bear-related impacts), as well as livestock

management (e.g. proactive measures prevent attacks by bears). To test for a possible relationship

between bear activity and conflicts, we first averaged predicted bear activity within the land

associated with each village. We then calculated Spearman rank correlations between average

predicted bear activity and: (i) the mean number of bear attacks on sheep in the last three years; and

(ii) perceived damage to orchards, crops and beehives. For the latter we used the proportion of

local people who stated that bears caused damage to local orchards, crops or beehives. To assess

whether certain herding techniques were more or less prone to attacks we used Spearman rank

correlations between number of bear attacks on sheep and (i) the number of sheep in the herd; (ii)

the number of sheep per guarding dog; (iii) total number of dogs, and (iv) the number of sheep per

shepherd.

Results

Predictive model of bear activity

Bear activity was recorded in 52% of the 630 transects. Cost-distance to the Carpathian Mountains

had the largest effect on bear activity (Table 4.1), which decreased with increasing cost-distance.

Biophysical variables also were strongly related to bear activity (Table 4.1). Bear activity was highest

in rugged terrain, near large forest patches with a low edge to interior ratio, and in areas with low

Chapter IV

82

pasture cover. The effects of distance to the village and local connectivity between forest patches

were weak and not significant (Table 4.1; Table S4.1).

Table 4.1 Model-averaged coefficient estimates (weighted mean ± weighted SE) of the bear activity model.

Ten separate models were initially calculated, and then a ten-fold cross-validation procedure was used. The

ten models were largely consistent in terms of which variables were significantly related to the response.

Number of times significant

in the ten separate models

Variable Estimate ± SE P < 0.05 P < 0.01

(Intercept) -1.66 ± 0.099 10 10

Number of anthills 0.04 ± 0.113 0 0

Number of anthills2 -0.29 ± 0.072 10 10

Time 0.58 ± 0.073 10 10

Time2 -0.24 ± 0.069 10 9

Distance to village 0.02 ± 0.069 0 0

Ruggedness 0.17 ± 0.059 10 7

Pasture cover -0.15 ± 0.069 8 3

Forest edge : forest area -0.17 ± 0.065 10 5

Betweenness centrality 0.04 ±0.076 0 0

Cost-distance to Carpathian Mountains -0.34 ±0.069 10 10

Predicted bear activity showed a gradient with proximity to the Carpathian Mountains but

was otherwise relatively homogenous throughout the region (Fig. 4.1c). We found no obvious

hotspots for bear activity inside the Natura 2000 area, but noted that predicted bear activity was

particularly high just north-east of (i.e. outside) the focal Natura 2000 area.

Human-bear conflicts

Predicted bear activity did not correlate with the average number of bear attacks on sheep reported

by shepherds (n = 73, rho = -0.1, p = 0.61). However, predicted bear activity was positively related

to the damage perceived by local villagers to orchards (n = 302, rho = 0.61, p < 0.001) and fields (n

= 302, rho = 0.5, p = 0.004), but not to beehives (n = 302, rho = 0.03, p = 0.86). All participants,

except one, stated that bear attacks on humans happened rarely or never.

The median herd size of sheep was 500, the median number of sheep per dog (typically

including one herding dog and several guarding dogs per herd) was 88, and the median number of

sheep per shepherd was 400. Of 34 described attacks in 23 villages, 56% occurred at night, 79% in

pastures, and 61% in shrubby places. In 69% of cases, less than three sheep were killed. Among the

shepherds who had suffered attacks, there was no correlation between the number of attacks and

the number of sheep per shepherd (rho = 0.13, p = 0.45), the number of sheep per dog (rho = 0.29,

p = 0.11), the total number of dogs (rho = 0.18, p = 0.11), or the total number of sheep (rho = 0.34,

p = 0.052). Many shepherds mentioned that having good dogs was the most important to avoid

sheep predation (pers. comm.).

The majority of shepherds perceived bear populations to be increasing over the past

decade, regardless of how many attacks they had suffered (Fig. 4.2a). However, shepherds who had


83

suffered more bear attacks perceived more strongly that bear attacks had increased over the past

decade (Fig. 4.2b). Despite these trends, approximately 50% of shepherds had neutral or positive

feelings towards bears (Fig. 4.2c).The percentage of shepherds who strongly disliked bears was

higher for shepherds who had suffered more attacks (Fig. 4.2c). Only six of 86 shepherds indicated

that their feelings towards bears had changed over the past decade. Over 50% of shepherds did not

support immediate killing of bears after attacks on livestock (indicated by ‘no’ and ‘depends’).

Shepherds who answered ‘depends’ supported the killing of a bear if livestock losses occurred

repeatedly by the same bear. Support for immediate killing (indicated by ‘yes’) was slightly higher

among shepherds who had suffered attacks than among those who had not (Fig. 4.2d).

Figure 4.2 Perceptions of shepherds regarding the brown bear. (A) trend in bear populations over the last

decade; (B) trend in attacks of bears on livestock over the last decade; (C) attitudes towards bears; and (D)

immediate elimination of bears after attacks on livestock. Cow herd = cow herders; no loss = shepherd with

no attacks by bears in the past three years; low loss = shepherd with fewer than ten attacks by bears in the

past three years; high loss = shepherds with ten or more attacks by bears in the past three years.

Ninety-three percent of shepherds found compensation for damage to livestock important.

However, 92% of shepherds agreed that compensation should be received only if appropriate

measures were taken by herders to protect livestock from carnivores. Interestingly, 90% of

shepherds were unaware of the (officially) existing compensation scheme. This finding, however,

should be interpreted cautiously because two shepherds indicated they had tried to receive

compensation, but were of the impression that no current compensation scheme was active. In

stark contrast, 85% of mayors, hunters and local councillors thought the compensation scheme was

Chapter IV

84

readily accessible; but only 43% of mayors, hunter and councillors believed that the majority of the

community was aware of the scheme.

Discussion

There is growing recognition of the need to conserve large carnivores outside protected areas (e.g.

Athreya et al. 2013; Schuette et al. 2013), however, insights from human-dominated landscapes are

still limited (e.g. Ghosal et al. 2013). Our approach drawing on both ecological and social data

suggests an apparent balance between humans and bears in Southern Transylvania. The main

factors contributing to this appeared to be: (i) the availability of large forest blocks connected to the

presumed source population of bears in the Carpathian Mountains; (ii) the use of traditional

livestock husbandry techniques to minimize damage from bears; and (iii) some tolerance among

shepherds to occasional conflict with bears. Unlike elsewhere, avoidance of human settlements by

bears (Posillico et al. 2004; Preatoni et al. 2005; Nellemann et al. 2007) and financial incentives

(Maclennan et al. 2009; Banerjee et al. 2013) appeared to play negligible roles in facilitating human-

bear coexistence.

Explaining coexistence

The Carpathian Mountains support the largest extant populations of the brown bear in Europe

(Zedrosser et al. 2001). Consistent with this, the most important predictor of bear activity was

proximity to the Carpathians. Higher bear activity was also found in rugged areas and near large

blocks of forest, which is most likely related to better shelter and den sites in rough terrain, and

potentially a wider variety of food resources (Nellemann et al. 2007; May et al. 2008; Güthlin et al.

2011). Local connectivity between forest patches, however, was unrelated to bear activity,

suggesting that forest patches are well-connected throughout the region. Indeed, forest cover in

our study area was close to the presumed 30% threshold below which the effects of habitat loss

may be exacerbated by isolation effects (Andrén 1994). Alternatively, local connectivity may be

related to shrub cover which was not accounted for in our analyses, and this may partly explain the

lack of statistical significance of local connectivity. In contrast to other studies (Posillico et al. 2004;

Preatoni et al. 2005; Nellemann et al. 2007), bears were not affected by distance to human

settlements. This may be because, at present, vehicle traffic does not increase strongly near

settlements, and major agricultural machinery is also relatively uncommon, although agricultural

intensification is likely in the future (Mikulcak et al. 2013). In addition, human presence per se may

not deter bears because local people are not a major threat, given that bears are protected by law

and hunting is prohibited. Finally, retaliation killing is probably uncommon in the study area,

though better knowledge on this would be useful to more fully understand human-bear

coexistence.

Interestingly, observed bear activity was not related to the frequency of attacks on sheep,

but was negatively related to perceived damage to orchards and fields. Unlike sheep, fields and


85

orchards are often guarded less carefully and most lack (effective) fences, suggesting that the

livestock husbandry techniques used by shepherds may effectively prevent bear attacks. The use of

guarding dogs and nightly confinement appears to reduce livestock attacks worldwide (Gehring et

al. 2010; Rigg et al. 2011). Indeed, most of the reported bear attacks occurred on the pasture, and

not in the well-guarded sheepfolds. Similar to observations in the Romanian mountains (Mertens &

Schneider 2005), we found no relationship between the frequency of bear attacks and the number

of sheep per dog. This is surprising, given the shepherds’ emphasis on the importance of good

guarding dogs. The lack of a relationship between attacked and the number of dogs could arise

from: (i) the use of an appropriately large number of dogs by the majority of shepherds; (ii) our

sample size being too small to detect a significant relationship; or (iii) our model for bear activity

only explaining bear distribution in spring and summer, while bear attacks occurred year-round.

The lack of a relationship between bear activity and frequency of attacks further suggests

that local conditions may be more important than actual carnivore densities in determining rates of

attack (Kaczensky 1999; Rigg et al. 2011). This supports the notion that human-carnivore

coexistence is possible, but knowledge of local conditions is necessary for effective, proactive

conflict management. In our case, relevant local conditions may include the quality of guarding

dogs and vigilance of shepherds, but also the prevalence of woody vegetation in pastures. Because

local conditions can, in principle, be managed, our findings indicate opportunities for proactive

conflict mitigation.

Public attitudes towards carnivores are typically most positive in areas without carnivores

(Kellert et al. 1996), or where people and carnivores have coexisted for a long time (Boitani 1995).

The long-term coexistence of bears and shepherds may have led to the acceptance of occasional

livestock loss by some shepherds, however, as observed elsewhere (Kaczensky et al. 2004),

shepherds that were heavily affected by bears were also less tolerant to livestock loss. Tolerance

could be further enhanced via compensation measures (Dickman et al. 2011) or the availability of

lethal measures to local authorities to take care of occasional ‘problem bears’ (Lescureux & Linnell

2010). At present, coexistence was not artificially upheld by economic incentives, suggesting that in

the absence of effective payment schemes, relatively cheap traditional, non-lethal methods can help

facilitate coexistence. Thus, the key to successful human-bear coexistence could lie in limiting

livestock losses to levels that are acceptable to a large proportion of the shepherd community,

while also establishing an understanding between local authorities and people that authorities will

increase efforts to prevent damage by bears. Developing such an understanding might be a difficult

challenge in Romania, where trust between local people and authorities is low due to historical

suppression, corruption and poverty (Hartel et al. 2014).

Chapter IV

86

Local management priorities

To facilitate the ongoing coexistence of bears and humans in southern Transylvania, conservation

measures should aim to maintain or improve: (i) connectivity between the foothills and the

Carpathian Mountains; (ii) availability of large blocks forest; and (iii) acceptance of occasional losses

to bears within the rural population. Maintaining regional connectivity and large forest blocks could

be challenging because of increasing pressure on forests from illegal logging activities (Knorn et al.

2012), and because of new major highways planned to cut through the study area. Highways can

negatively impact bear populations by causing habitat fragmentation and increasing the risk of

collisions with vehicles (Kaczensky et al. 2003; Karamanlidis et al. 2012). Wildlife crossing

structures could partly counteract these impacts, but research is needed to identify suitable

locations for such structures (Clevenger & Waltho 2001). Furthermore, there is the danger that

conservation efforts will focus on the protected Natura 2000 area, which does not extend all the

way to the Carpathian Mountains, and does not capture all of the most important areas for the

brown bear (Fig. 4.1c).

Compensation payments are often used to increase tolerance levels of people negatively

affected by carnivores (Dickman et al. 2011). In Romania, damage caused by protected species

should be compensated through the central public authority for environmental protection (law

407/2006). Yet, most shepherds indicated they found compensation payments important but did

not know about the existence of the compensation scheme. Many officials dealing with

compensation thought the scheme was still relatively unknown, but overall, despite a complicated

application process (Mertens & Promberger 2001), they believed that the scheme was readily

accessible. This suggests that the compensation payments need to be more transparent and

accessible to local people. Importantly, monetary compensation is not the only plausible policy

option. Proactive payments for preventive measures may be more successful in improving

conditions for coexistence (Swenson & Andrén 2005), and subsidizing electric fences along forest

edges and around sheepfolds could help reduce sheep predation. Moreover, given widespread

mistrust in authorities, bottom-up compensation payments organized by local groups could be

more effective. For example, contributions to a local livestock insurance program (Mishra et al.

2003) or replacements of lost livestock from a communal compensation herd may be worth

considering.

Limitations

The proportion of destroyed anthills to total anthills turned out to be a low-cost and pragmatic, but

evidently useful, index of bear activity. However, this metric has several limitations. First, bear use

of anthills in a given pasture might be influenced by overall levels of anthill availability and by other

pasture characteristics. Our large sample size and stratification of transects should minimize

systematic biases caused by general pasture characteristics, and the availability of anthills was

accounted for in the model. Therefore, we are confident that our indicator measured actual bear


87

activity rather than bear exploitation of anthills. Second, ant larvae are a seasonal food source, and

thus, other variables may explain bear distribution patterns in other seasons. The seasonal nature of

our surveys could partly explain the lack of a correlation between bear activity and self-reported

attacks of bears on sheep, which occurred year round. That said, we did find significant correlations

between bear activity and perceived damage to orchards and fields, suggesting a certain level of

cross-validation of methods – by definition, it is highly unlikely that a statistical significant

correlation between our activity index and perceived damage would have arisen by chance if there

was in fact no relationship between these two variables. Third, our data on damage caused by bears

was based on the perceptions of shepherds and local people, and was not validated by official

damage reports. Interview data therefore should be interpreted with care. While these limitations

should be kept in mind, we do not believe they fundamentally undermine the validity of our overall

findings.

Future research and conclusion

Although coexistence of humans and carnivores is a socially desired goal in many landscapes

around the world, most research takes place in protected areas with few people (Ghosal et al.

2013). Our interdisciplinary approach demonstrated the usefulness of combining ecological and

social data to highlight conservation priorities in carnivore conservation. Arguably, framing

carnivore conservation in a social-ecological context would also be useful in other human-

dominated landscapes.

Our study indicates that coexistence of humans and carnivores is possible, even without

direct economic incentives. Continuous coexistence with large carnivores appears to foster the

development of management tools and attitudes that effectively reduce conflicts. Nevertheless, this

shared history of relationships between humans and bears has been eroded in many regions

worldwide. Thus, a key challenge for settings with a broken history of human-carnivore co-

occurrence is to reinstate both practices and attitudes that facilitate coexistence. While a history of

continuous coexistence cannot be re-created in places where carnivores have been extirpated, it is

noteworthy that in areas were carnivores are slowly re-colonizing landscapes, initially negative

attitudes can become more neutral as people once again become accustomed to living with

carnivores (Majić & Bath 2010).

Although our study indicates that coexistence is possible, the functional mechanisms

facilitating this remain poorly understood. One recently discussed mechanism is that of behavioural

adjustment on behalf of the carnivores, who may adjust temporal activity patterns (Martin et al.

2010; Carter et al. 2012b). Our study suggests that ‘behavioural’ mechanisms on behalf of people –

that is, social mechanisms – also deserve more attention. Social mechanisms underpinning human-

carnivore coexistence are acknowledged by several authors (e.g. Carter et al. 2012a; Athreya et al.

2013), but still remain poorly accounted for in many studies on human-carnivore conflicts, as well

as in many mitigation programs (Dickman 2010; Glikman & Frank 2011). Future research should

Chapter IV

88

investigate the drivers of human attitudes towards carnivores, which may vary substantially in

relation to the behaviour and ecology of the species in question, the prevalence of traditional

ecological knowledge, dominant cultural values and beliefs, differences in social equity and

distribution of carnivore impacts, and political context (Dickman 2010; Lescureux & Linnell 2010).

Acknowledgments

We thank Z. Aczél-Fridrich, R. Bigonneau, M. Brix, K. Delille, L. A. Dénes, A. Fletcher, T. Fülöp,

F. Helissandre, C. Klein, I. Kósa, A. Krieg, C. Moga, N. Moraru, M. Roellig, B. Scheele, J.

Schultner, A. Szapanyos, L. Teixeira, and J. Tinnemans for field efforts and data entry. We thank

the shepherds and local people, and O. Bodin for advice on connectivity analyses. We thank two

anonymous reviewers and the handling editor for constructive suggestions on a previous draft of

this manuscript. The project was funded by a grant of the International Association for Bear

Research and Management to ID and a Sofja Kovalevskaja Award by the Alexander von Humboldt

Foundation to JF. Part of the data collection was supported by the European Regional

Development Fund (ERDF) through Sectorial Operational Plan - Environment (POS Mediu)

under the project ‘For nature and local communities - the basis for a Natura 2000 integrated

management in the Hârtibaciu - Târnava Mare - Olt area’.


89

Supplementary Material for Chapter IV

Text S4.1. Additional information on the study area and bear activity model.

Study area

The overall character of Transylvanian villages and the surrounding land has changed relatively

little for decades (Wilkie 2001; Akeroyd & Page 2007). The on-going practice of low-intensity semi-

subsistence farming has maintained exceptional levels of biodiversity (Cremene et al. 2005;

Akeroyd & Page 2007). The study area had approximately equal amounts of the main land-use

types forests (28%), pasture (24%), and arable land (37%), and altitudes ranged between 230 and

1100 m above sea level. Most forests are deciduous and dominated by hornbeam (Carpinus betulus),

oak (Quercus sp.), and beech (Fagus sylvatica). Pastures occur on the slopes and are grazed by sheep

(dominant livestock), goats and cattle. Forest edges usually abut pastures. Low-intensity small scale

farming occupies the valleys. Recently, one of the largest sets of non-mountainous Natura 2000

sites in the EU has been established in the study area, including both Sites of Community

Importance (SCI, Habitats Directive), and Special Protection Areas (SPA, Birds Directive; Fig.

4.1b).

Details on the variables used to model bear activity

Distance to the nearest settlement was measured as the distance from the midpoint of a given

transect to the edge of the nearest village (min = 30 m, max = 4368.7 m, median = 1505.8 m) and

was included because human activities often negatively affect bear presence (Nellemann et al. 2007;

Fernández et al. 2012). We did not include the distance to roads nor human population density,

because there are few big roads, and population density was relatively uniform across the region.

The biophysical variables were calculated within a radius of 1000 m from the midpoint of a

given transect, which is slightly larger than the straight line daily distance travelled by brown bears

in Croatia (Huber & Roth 1993). Variables included: (i) terrain ruggedness calculated as the

standard deviation of the altitude (Advanced Spaceborne Thermal Emission and Reflection

Radiometer (ASTER) Global Digital Elevation Model Version 2 (GDEM V2)), because more

rugged areas may offer more and better resting and sheltering sites (Naves et al. 2003); (ii) percent

pasture cover as a general indicator of the extent to which the landscape was used for livestock

grazing; and (iii) forest edge to forest interior ratio as an indication of the shapes and sizes of forest

patches (patches with lower edge to interior ratios were both rounder and larger). We first

calculated forest cover for the entire study area within a moving window with a radius of 1000 m.

Second, we extracted all forest edges as polylines in ArcGIS, and converted them into pixels of 10

x 10 m to create a measure of forest-edge cover. Third, we used these two variables to calculate the

forest edge to forest interior ration. Hence, no artificial forest edges were created by the circle itself.

Forest and pasture variables were derived from the Corine Land Cover digital map 2006 (CLC). All

variables were calculated and mapped for the entire study area with a resolution of 10 m x 10 m.

Chapter IV

90

Due to the large home range of the brown bear (58 km2 – 128 km2; Huber & Roth 1993),

habitat characteristics probably influence bear activity beyond a 1000 m radius. Therefore, we

included a measure of local connectivity between forest patches, indicated by the betweenness

centrality of the forest patch k that was closest to a given transect. Betweenness centrality is the

sum of all shortest pathways between all pairs of patches that go through patch k: it measures how

much patch k is involved in potential movements between other pairs of patches by serving as an

intermediate stepping stone patch (Bodin & Saura 2010). Betweenness centrality was calculated

with Conefor Sensinode 2.2 (Saura & Torne 2009), between the edges of individual forest patches

using a dispersal distance of 2000 m (Huber & Roth 1993).

The presumed source population of brown bears resides in the Carpathian Mountains and

thus proximity to the mountains is likely to affect bear activity. We used the cost-distance from the

midpoint of a given transect to the mountains in our models, which we calculated using the cost-

distance analysis tool in ArcGIS 10.0 at a resolution of 10 m x 10 m. Expert opinions are widely

used in animal habitat suitability models and corridor analyses (e.g. Rathore et al. 2012), and

therefore we asked two local bear experts in Romania to weight the different main land-use types

of the Corine Land Cover Map 2006 on a scale from 1-10 with regard to their effects on bear

movement (Table S4.2). Expert 1 has approximately 20 years of working experience with bears in

Transylvania including basic research on bears, monitoring movement activities with radio-collars,

conflict mitigation and non-invasive genetic monitoring. Expert 2 has approximately 10 years of

working experience with bears in Transylvania including basic research on bears and conflict

mitigation. Based on the weights provided by these experts, we developed two alternative cost

surfaces for the cost-distance analysis in ArcGIS. To decide which cost-distance matrix performed

best we compared three generalized linear models with a quasi-binomial error structure and the

proportion of destroyed anthills as response variable. For this procedure, we used only those 59

transects at an approximately constant distance of 15-20 km from the Carpathians. The three

different models included the same variables as the model used to predict bear activity; however,

they differed in their cost-distance measures. Model 1 and 2 each reflected a cost-distance measure

according to either of the experts, while model 3 did not include a cost-distance measure. In the

latter case, we could have used ‘raw distance’ as an alternative (with no weightings for land cover

types), but we did not deem this meaningful because raw distance to the mountains was held

approximately constant during this calibration. The cost-distance measure associated with the

model that explained most deviance was used in further analyses (Table S4.3).


91

Details on statistical modelling

Prior to modelling we log-transformed the number of anthills, distance to village, and forest edge

to interior ratio; we confirmed that variables were not correlated (allowing only Pearson's r < 0.5);

and we standardized all variables to zero mean and unit variance. We pooled transect data from the

30 villages with data from the Natura 2000 site. We used principal components analysis to confirm

that the environmental variables spanned similar gradients in these two datasets to ensure they

could be meaningfully combined.

All analyses were performed with the proportion of destroyed vs. total number of anthills as

the response variable within generalized linear models (GLM) with a quasi-binomial error structure

to account for overdispersion. We first included all data into a global model to check model

assumptions and test for spatial autocorrelation. We found spatial autocorrelation in the first five

bins using an increment of 4000 m, however, the correlation was low (first bin correlation = 0.13, p

< 0.001). To correct for this (relatively mild) autocorrelation, we ran a mixed effects model

(glmmPQL) with an exponential spatial correlation structure, using the R packages ‘MASS’.

Although the mixed model diminished spatial autocorrelation of the data, the model estimates were

very similar to the estimates in the quasi-binomial GLM (Table S4.4). Therefore, we chose to

present the modelling results from the simpler quasi-binomial generalized linear model, and also

used it to predict bear activity for the whole study area.

The final GLM was obtained by using model averaging with a 10-fold cross-validation

(Fielding & Bell 1997). This means we randomly divided our data into ten equally-sized groups,

successively left out one tenth of the data to calibrate the models, and predicted and tested bear

activity for the omitted tenth. Validation was done by relating the predicted activity (obtained from

a model using nine tenths of the data) with the activity observed in the remaining tenth via a

binomial GLM. We calculated the amount of deviance explained as a measure of predictive

performance of each of the ten models. Finally, we averaged the parameter estimates over the 10

models, weighting each model by its predictive performance. The averaged parameter estimates

then were used to predict bear activity. Table S4.1 shows the estimates and explained deviance used

for each of the validation steps.

We predicted bear activity for all 800 m segments (n = 7536) of forest-pasture interface in the

entire study area. We first extracted all forest-pasture interfaces as a polyline from the Corine Land

Cover Map 2006 using ArcGIS 10.0. We divided all forest-pasture polylines into segments of 800

m, because this was equal to the length of our transects. Then we extracted the middle point from

each segment. We used these points to extract the values for all variables used in the bear activity

model to predict bear activity across the study area. To extract the variables we used the same maps

(resolution 10 m x 10 m) we created for our bear activity model described above. The total number

of anthills presumed to exist at any given forest edge was set to the mean observed, and time was

set to the time when observed bear activity peaked. All statistical analysis were implemented in the

‘R’ environment (R Development Core Team 2013).

Chapter IV

92

Ethical considerations for the interviews

Prior to all interviews, informed voluntary consent was obtained orally by all interviewees.

Interviewees were provided with information about the research project before the interview.

Interviewees had the opportunity to withdraw from interview at any time if they felt uncomfortable

with the questions. The privacy of the interviewees was protected by keeping them anonymous.

Data were not of a personal nature, and no interviewees expressed concerns.


93

Text S4.2. The questionnaire for the personal interviews with 86 shepherds (original in Romanian).

With this questionnaire we would like to learn more about the problems between people and bears and wolves.

Date: Village: GPS point: Please indicate how often you encounter:

Never or very rarely

About once a year

Several times a year

About once a month

About once a week

Bears

Wolves

The following questions will ask about your experience with damage by bears and wolves

1. Have you ever lost livestock due to bears? a) yes b) no

2. How many times did bears attack your livestock in the last 3 years:

3. Have you ever lost livestock due to wolves?

a) yes b) no

4. How many times did wolves attack your livestock in the last 3 years: If you have ever suffered an attack by bears and/or wolves answer the following questions regarding the last attack for each species. Bears

5. Date of the attack o Year: o Month:

6. Which livestock was present and approximately how many of each?

o Sheep: present: a) yes b) no c) number: o Cows: present: a) yes b) no c) number: o Goat: present: a) yes b) no c) number: o Horse: present: a) yes b) no c) number: o Donkey: present: a) yes b) no c) number:

7. What were the losses?

o Animal species 1: o Loss in numbers:



8. When did the attack happen? a) day b) night

9. Where did the attack happen?

a) in the sheepfold c) in the forest b) on the pasture d) other, namely:

Chapter IV

94

10. Did the attack happen in the location of this sheepfold? a) yes b) no, in another location, namely:

11. What did the surroundings look like?

a) open landscape c) other, namely: b) shrubby landscape

12. How many shepherds were present?

13. What was the behavior of the shepherds present? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

14. How many dogs were present?

15. What was the behavior of the dogs present?

______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

16. Would you like to add anything else that could explain why the attack was successful or

unsuccessful? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

Wolves

17. Date of the attack o Year: o Month:

18. Which livestock was present and approximately how many of each?

o Sheep: present: a) yes b) no c) number: o Cows: present: a) yes b) no c) number: o Goat: present: a) yes b) no c) number: o Horse: present: a) yes b) no c) number: o Donkey: present: a) yes b) no c) number:

19. What were the losses? o Animal species 1: o Loss in numbers:



20. When did the attack happen? a) day b) night


95

21. Where did the attack happen?

c) in the sheepfold c) in the forest d) on the pasture d) other, namely:

22. Did the attack happen in the location of this sheepfold? b) yes b) no, in another location, namely:

23. What did the surrounding look like? a) open landscape c) other, namely: b) shrubby landscape

24. How many shepherds were present?

25. What was the behavior of the shepherds present? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

26. How many dogs were present?

27. What was the behavior of the dogs present?

______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

28. Would you like to add anything else that could explain why the attack was successful or

unsuccessful? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

This section will ask for your opinion about a list of statements on bears and wolves. In your opinion:

29. Bear populations over the last 10 years have: a) increased b) remained stable c) decreased d) don’t know

30. Wolf populations over the last 10 years have: a) increased b) remained stable c) decreased d) don’t know

31. Sheep populations in my region over the last 10 years have: a) increased b) remained stable c) decreased d) don’t know

32. Cow populations in my region over the last 10 years have: a) increased b) remained stable c) decreased d) don’t know

Chapter IV

96

33. Bear attacks on livestock over the last 10 years have:

a) increased b) remained stable c) decreased d) don’t know

34. Wolf attacks on livestock over the last 10 years have: a) increased b) remained stable c) decreased d) don’t know

35. Which of the following describes your feeling towards bears?

a) do not like at all b) do not like c) neutral d) I rather like e) I like very much

36. Has your attitude towards bears changed over the last 10 years? a) yes b) no

37. If yes, have you become a) more negative b) more positive

38. Why has your attitude changed?

______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

39. Which of the following describes your feeling towards wolves?

a) do not like at all b) do not like c) neutral d) I rather like e) I like very much

40. Has your attitude towards wolves changed over the last 10 years? a) yes b) no

41. If yes, have you become a) more negative b) more positive

42. Why has your attitude changed?

______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

The following questions are about bear attacks on humans.

43. Bear attacks on humans happen: a) never b) rarely c) regularly d) very often

44. Bears only attack humans when they feel threatened by humans: a) yes b) no

45. I am afraid to be attacked by a bear: a) yes b) no

The following questions are about compensation schemes.

46. It is important to receive compensation for livestock losses:

a) yes b) no


97

47. Compensation should only be paid when livestock is properly guarded, e.g. by dogs or

fences

a) yes b) no

48. Bears and wolves should be eliminated after damaging livestock:

a) yes b) no

49. Compensation is currently being paid for livestock losses:

a) yes b) no c) don’t know

48. Have you ever asked for compensation for livestock losses: a) yes; b) no

If yes:

a. Was it easy to receive the compensation: a) yes b) no

b. What did you need to do to receive compensation:

______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

The following questions are about you.

48. Gender: a) M b) F

49. Age:______years

50. Ethnicity: a) Romanian b) Hungarian c) Roma d) Saxon e) Other, namely:

51. Which animals do you have in your camp and how many of each? o Sheep number: o Cows number: o Goat number: o Horse number: o Donkey number: o Dogs number:

Thank you very much for your participation!

Chapter IV

98

Text S4.3. The questionnaire used for interviews with 302 local people, 15 hunters and 20 local

councillors (original in Romanian).

With this questionnaire we would like to learn more about the problems between people and bears and wolves.

Date: Village: The following questions will ask about your experience with damage by bears and wolves. Please indicate how often you encounter:

Never or very rarely

About once a year

Several times a year

About once a month

About once a week

Bears

Wolves

Please indicate if you agree or disagree.

1. My village suffers great damage to fields by bears? a) yes b) no

2. My village suffers great damage to orchards by bears? a) yes b) no

3. My village suffers great damage to beehives by bears? a) yes b) no

4. My village suffers great damage to livestock by bears? a) yes b) no

5. My village suffers great damage to livestock by wolves? a) yes b) no

The following questions are about bear attacks on humans.

6. Bear attacks on humans happen: a) never b) rarely c) regularly d) very often

7. Bears only attack humans when they feel threatened by humans: a) yes b) no

8. I am afraid to be attacked by a bear: a) yes b) no

9. Do you have anything you would like to add regarding your opinion about bears or

wolves? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

The following questions are about compensation schemes and should only be filled in for members of the local council and hunters.

10. It is important that compensation is paid for livestock losses and/or damage to crops by wolves and bears a) yes b) no c) don’t know

11. In my village compensation for damage to livestock by bears and/or wolves is frequently

requested a) yes b) no c) don’t know


99

12. In my village compensation for damage to crops by bears is often requested

a) yes b)no c) don’t know

13. Most people in my village are aware of the existing compensation scheme for damage caused by bears and wolves a) yes b)no c) don’t know

14. Most applications for compensations are successful

a) yes b)no c) don’t know

15. Could you briefly describe the steps that need to be taken to receive compensation for damage inflicted by bears and wolves? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

16. Do you think the procedure you described is accessible to everybody? a) yes b) no c) don’t know



18. Age:______years

19. Profession

20. Ethnicity: a) Romanian b) Hungarian c) Roma d)Saxon e) Other, namely:

21. Do you have: i) Land with crops a) yes b) no ii) Orchards a) yes b) no iii) Beehives a) yes b) no iv) Livestock a) yes b) no


Chapter IV

100

Text S4.4. The questionnaire used for interviews with 22 mayors or vice-mayors (original in

Romanian).

Date: Commune: In this questionnaire we would like to know about the problems between humans and bears/wolves in your commune.

The following questions are about damage caused by bears and wolves. Please indicate if you agree or disagree.

22. My commune suffers great damage to fields by bears? a) yes b) no

23. My commune suffers great damage to orchards by bears? a) yes b) no

24. My commune suffers great damage to beehives by bears? a) yes b) no

25. My commune suffers great damage to livestock by bears? a) yes b) no

26. My commune suffers great damage to livestock by wolves? a) yes b) no 27. Bear attacks on humans in my commune happen:

a) rarely b) regularly c) very often d) never

The following questions are about compensation schemes.

I. It is important that compensation is paid for livestock losses and/or damage to crops by wolves and bears a) yes b) no c) don’t know

II. In my village compensation for damage to livestock by bears and/or wolves is frequently

requested a) yes b) no c) don’t know

III. In my village compensation for damage to crops by bears is often requested a) yes b)no c) don’t know

IV. Most people in my village are aware of the existing compensation scheme for damage caused by bears and wolves a) yes b)no c) don’t know

V. Most applications for compensations are successful

a) yes b)no c) don’t know VI. Could you briefly describe the steps that need to be taken to receive compensation for

damage inflicted by bears and wolves : ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________

VII. Do you think the procedure you described is accessible to everybody? a) yes b) no c) don’t know


101

VIII. Do you have anything you would like to add regerading your opinion about bears or wolves? ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________ ______________________________________________________________________



9. Age:______years

10. Ethnicity: a) Romanian b) Hungarian c) Roma d)Saxon e) Other, namely:


Chapter IV

102

Table S4.1 Model coefficient estimates (± SE) of the ten models used in the 10-fold cross-

validation and their explained deviance. An asterisk (*) indicates significant variables (p < 0.05).

Explained deviance was calculated as the amount of deviance explained from a binomial GLM

relating the modelled and observed bear activity.

Variable Model 1 Model 2 Model 3 Model 4 Model 5

(Intercept) -1.74 ± 0.101* -1.68 ± 0.100* -1.63 ± 0.100* -1.72 ± 0.098* -1.63 ± 0.101*

Number of anthills 0.04 ± 0.115 0.10 ± 0.118 0.01 ± 0.111 0.07 ± 0.113 0.02 ± 0.116

Number of anthills2 -0.28 ±0.071* -0.35 ± 0.078* -0.25 ± 0.081* -0.28 ± 0.069* -0.29 ± 0.073*

Time 0.55 ± 0.071* 0.56 ± 0.071* 0.56 ± 0.071* 0.58 ± 0.071* 0.61 ± 0.077*

Time2 -0.17 ± 0.069* -0.23 ± 0.067* -0.27 ± 0.069* -0.23 ± 0.069* -0.28 ± 0.072*

Distance to village -0.02 ± 0.068 0.03 ± 0.070 0.06 ± 0.069 0.03 ± 0.069 0.02 ± 0.072

Ruggedness 0.15 ± 0.059* 0.21 ± 0.057* 0.18 ± 0.057* 0.16 ± 0.057* 0.19 ± 0.061*

Pasture cover -0.14 ± 0.069* -0.17 ± 0.070* -0.20 ± 0.070* -0.15 ± 0.068* -0.15 ± 0.070*

Forest edge : forest

area

-0.18 ± 0.065* -0.20 ± 0.064* -0.15 ± 0.064* -0.20 ± 0.066* -0.16 ± 0.066*

Betweenness centrality 0.02 ± 0.076 0.08 ± 0.077 0.00 ± 0.078 0.03 ± 0.075 0.03 ± 0.077

Cost-distance to

Carpathian Mountains

-0.29 ± 0.069* -0.39 ± 0.070* -0.38 ± 0.069* -0.32 ± 0.070* -0.34 ± 0.069*

Explained deviance 0.35 0.12 0.19 0.24 0.27

Model 6 Model 7 Model 8 Model 9 Model 10

(Intercept) -1.64 ± 0.100* -1.61 ±0.093* -1.62 ± 0.097* -1.67 ± 0.100* -1.65 ± 0.100*

Number of anthills 0.03 ± 0.112 0.03 ± 0.107 0.01 ± 0.110 0.09 ± 0.114 0.06 ± 0.115

Number of anthills2 -0.30 ± 0.071* -0.30 ± 0.067* -0.28 ± 0.069* -0.30 ± 0.071* -0.29 ± 0.071*

Time 0.56 ± 0.073* 0.67 ± 0.073* 0.57 ± 0.071* 0.57 ± 0.073* 0.60 ± 0.075*

Time2 -0.23 ± 0.070* -0.32 ± 0.068* -0.26 ± 0.068* -0.24 ± 0.068* -0.25 ± 0.070*

Distance to village 0.02 ± 0.070 -0.02 ± 0.068 0.07 ± 0.069 -0.01 ± 0.070 0.03 ± 0.070

Ruggedness 0.17 ± 0.060* 0.14 ± 0.057* 0.14 ± 0.058* 0.17 ± 0.057* 0.19 ± 0.059*

Pasture cover -0.15 ± 0.068* -0.18 ± 0.065* -0.19 ± 0.070* -0.13 ± 0.068 -0.12 ± 0.068

Forest edge : forest

area

-0.16 ± 0.066* -0.16 ± 0.063* -0.13 ± 0.065* -0.20 ± 0.067* -0.18 ± 0.065*

Betweenness centrality 0.06 ± 0.077 0.01 ± 0.074 0.06 ± 0.074 0.05 ± 0.074 0.05 ± 0.076

Cost-distance to

Carpathian Mountains

-0.36 ± 0.069* -0.39 ± 0.065* -0.33 ± 0.068* -0.35 ± 0.068* -0.33 ± 0.071*

Explained deviance 0.28 0.02 0.39 0.19 0.36


103

Table S4.2 Weights assigned to different land-uses regarding bear movements by the two local

bear experts. The weights spanned a gradient between 1 and 10, with 10 being difficult for bear

movement and 1 being easy. The weights by Expert 1 were ultimately used in the activity model

because they provided a better fit to the data (see Table S4.3).

Land cover type Expert 1 Expert 2

Human settlement/industrial units 8 10

Agricultural land 3 8

Pasture 4 7

Forest 1 1

Woodland/scrubland 1 2

Major rivers 3 5

Orchards 2 4

Major roads 6 5

Table S4.3 Model coefficient estimates (± SE) of the three models used to choose the most

appropriate cost-distance measure to be used in the main bear activity model. These models were

constructed on the basis of the 59 transects located at 15-20 km from the Carpathians (see Figure

4.1a).

Expert 1 Expert 2 No cost-distance

Variable Estimate ± SE p Estimate ± SE p Estimate ± SE p

(Intercept) -3.83 ± 0.846 <0.001 -2.88 ± 0.706 <0.001 -2.20 ± 0.491 <0.001

Number of anthills -0.37 ± 0.274 0.184 -0.32 ± 0.282 0.267 -0.29 ± 0.283 0.317

Number of anthills2 -0.13 ± 0.139 0.370 -0.08 ± 0.144 0.572 -0.09 ± 0.146 0.561

Time 0.89 ± 1.950 0.651 0.69 ± 1.989 0.729 0.30 ± 1.910 0.877

Time2 1.30 ± 1.693 0.445 0.97 ± 1.704 0.571 0.41 ± 1.596 0.780

Distance to village -0.33 ± 0.158 0.044 -0.28 ± 0.168 0.102 -0.22 ± 0.161 0.171

Ruggedness -0.15 ± 0.227 0.502 -0.02 ± 0.243 0.918 0.21 ± 0.166 0.205

Pasture cover 0.16 ± 0.24 0.516 0.14 ± 0.253 0.582 0.09 ± 0.245 0.729

Forest edge : forest area -0.31 ± 0.174 0.085 -0.36 ± 0.192 0.064 -0.29 ± 0.177 0.103

Betweenness centrality 0.15 ± 0.101 0.145 0.07 ± 0.124 0.581 0.17 ± 0.101 0.094

Cost-distance to Carpathian Mountains

-1.81 ± 0.764 0.022 -0.70 ± 0.512 0.175 - / - - / -

Proportion of deviance explained by the model

0.33 0.28 0.25

Chapter IV

104

Table S4.4 Model coefficient estimates (± SE) of the global Generalized Linear Model (GLM) and

the alternative Generalized Linear Mixed Model (GLMM) that included a correlation function to

correct for spatial autocorrelation. Because parameter estimates were very similar, the simpler GLM

was used as the final bear activity model.

GLM GLMM

Variable Estimate ± SE p Estimate ± SE p

(Intercept) -1.66 ± 0.09 <0.001 -1.71 ± 0.106 <0.001

Number of anthills 0.04 ± 0.11 0.675 0.02 ± 0.11 0.861

Number of anthills2 -0.29 ± 0.068 <0.001 -0.26 ± 0.068 <0.001

Time 0.58 ± 0.069 <0.001 0.58 ± 0.074 <0.001

Time2 -0.25 ± 0.065 <0.001 -0.26 ± 0.069 <0.001

Distance to village 0.02 ± 0.066 0.765 0.04 ± 0.072 0.536

Ruggedness 0.17 ± 0.055 0.002 0.15 ± 0.059 0.010

Pasture cover -0.16 ± 0.065 0.016 -0.18 ± 0.069 0.010

Forest edge : forest area -0.17 ± 0.062 0.006 -0.16 ± 0.066 0.017

Betweenness centrality 0.04 ± 0.072 0.580 0.07 ± 0.078 0.403

Cost-distance to Carpathian Mountains -0.35 ± 0.065 <0.001 -0.38 ± 0.073 <0.001

105

Chapter V

106

107

Chapter V

Incorporating anthropogenic effects into trophic ecology:

Predator-prey interactions in a human-dominated landscape

Ine Dorresteijn, Jannik Schultner, Euan G. Ritchie, Dale G. Nimmo, Jan Hanspach, Tobias

Kuemmerle, Laura Kehoe, Joern Fischer

In revision in Proceedings of the Royal Society B: Biological Sciences

‘What escapes the eye…is a

much more insidious kind of

extinction: the extinction of

ecological interactions.’

–Daniel H. Janzen

‘The last word in ignorance is the man who says of an

animal or plant: 'What good is it?’ – Aldo Leopold

Forest cover

Cervus elaphus

Pasture cover

Canis lupus Ursus arctos Canis familiaris

Capreolus capreolus

Homo sapiens

Vulpes vulpes

Chapter V

108

Human influence on trophic interactions

109

Abstract

Apex predators perform important functions that regulate ecosystems worldwide. However, little is

known about how ecosystem regulation by predators is influenced by human activities. In

particular, how important are top-down effects of predators relative to direct and indirect human-

mediated bottom-up and top-down processes? Combining data on species’ encounter rates from

camera traps and hunting records, we aimed to understand and quantify the relative effects of top-

down and bottom-up processes in shaping predator and prey distributions in a human-dominated

landscape in Transylvania, Romania. By global standards this system is diverse, including apex

predators (brown bear and wolf), mesopredators (red fox) and large herbivores (roe and red deer).

Furthermore, humans and free-ranging dogs represent additional predators in the system. Using

structural equation modelling we found that apex predators maintained their ecological role by

suppressing lower trophic levels, especially herbivores. However, direct and indirect human top-

down effects at multiple trophic levels affected the ecosystem more strongly, influencing species at

all trophic levels. Our study highlights the need to explicitly embed humans and their influences

within trophic cascade theory. This will greatly expand our understanding of species interactions

and ecosystem dynamics in human-modified landscapes, which compose the majority of Earth’s

terrestrial surface.

Chapter V

110

Introduction

There is increasing recognition of the critical role apex predators play in structuring ecosystems

globally (Estes et al. 2011; Ripple et al. 2014). They do so by killing or instilling fear in competitors

and prey (Creel & Christianson 2008; Ritchie & Johnson 2009), thereby inducing trophic cascades

that flow through entire ecosystems (Letnic et al. 2009). Despite 40% of the Earth’s terrestrial

surface being dominated by agriculture (Foley et al. 2005) and human’s effects permeating into

more natural areas (Sanderson et al. 2002), most research on trophic cascades has focused on

relatively intact wilderness and conservation reserves. However, top-down processes (i.e. the

structuring of the ecosystem by high trophic levels) and bottom-up processes (i.e. control through

productivity and low trophic levels) in largely pristine areas may differ substantially from those in

human-dominated landscapes. Thus, a key question remains: what role do humans play in the

trophic networks of modified ecosystems (Sergio et al. 2014)? Answering this question is important

for a number of reasons. Firstly, many large carnivore populations exist outside protected areas and

are embedded within human-dominated landscapes (Crooks et al. 2011; Carter et al. 2012b; Vanak

et al. 2013). Secondly, there is increased focus on using large carnivores in the context of ecosystem

restoration (Ritchie et al. 2012). And thirdly, in parts of the world such as Europe and North

America, large carnivores are returning through active rewilding efforts and conservation initiatives

(Enserink & Vogel 2006; Navarro & Pereira 2012; Morell 2013). Together this highlights the urgent

need to better understand relationships between apex predators, people, and ecosystem

components in human-dominated landscapes.

In terrestrial ecosystems, apex predators have been linked to two major trophic cascades.

First, apex predators limit herbivores through direct predation and behaviourally mediated changes

in habitat use, thereby promoting vegetation growth (i.e. tri-trophic cascades; Beschta & Ripple

2009; Ripple & Beschta 2012b; Kuijper et al. 2013). Second, apex predators limit smaller predators

through interference competition, including in its most extreme form, intraguild predation (Polis &

Holt 1992; Palomares & Caro 1999; Brook et al. 2012). Mesopredator suppression by apex

predators can thereby increase the abundance of small mammals and birds (i.e. mesopredator

cascades; Crooks & Soulé 1999; Ritchie & Johnson 2009). Such cascading effects could differ in

human-dominated landscapes in at least two main ways. For example, humans influence species

abundances through bottom up processes such as land-use, agriculture, and forestry, which may

translate into a wide range of changes in ecosystem properties and functions (Foley et al. 2005),

including ecosystem productivity (Haberl et al. 2007), or food and habitat availability (Muhly et al.

2013). Such changes in productivity can significantly modify predator-prey relationships (i.e., the

Ecosystem Exploitation Hypothesis; Oksanen & Oksanen 2000). Also, humans directly (e.g.

harvesting of both predators and prey; Barnosky et al. 2004) or indirectly (e.g. by creating an

anthropogenic landscape of fear; Ordiz et al. 2013) affect top-down processes and it remains

unclear if apex predators can achieve high enough densities outside wilderness areas and protected


111

areas to be ‘ecologically effective’ (Soulé et al. 2003; Letnic et al. 2012; Mech 2012; Ordiz et al.

2013). These latter effects in particular have been overlooked in literature on trophic cascades.

Useful insights into the role of humans could be gained by studying ecosystems in which

both humans and carnivores have coexisted for extended periods. Traditional farming regions in

Romania form an ideal system in this respect. The forests surrounding the villages cover a third of

the area and are well-connected (Dorresteijn et al. 2014). The heterogeneous landscape harbors

cervid herbivore and mesopredator populations, but also relatively high densities of brown bears

(Ursus arctos) and lower densities of the grey wolf (Canis lupus). The use of free-ranging large-bodied

livestock guard dogs (Canis familiaris) to protect livestock against carnivores adds a third non-

human, predator to the system.

Wolves are the most important cervid predator in the Northern Hemisphere (Peterson et

al. 2003), and are involved in both tri-trophic and mesopredator cascades (Berger et al. 2008; Ripple

& Beschta 2012b). Bears are omnivorous and may not be able to limit herbivore populations alone

(Ripple & Beschta 2012a), and their effects on mesopredators remain unclear. However, bears can

limit cervid densities in combination with wolves, and their predation on cervid calves may affect

the recruitment of juveniles (Berger et al. 2001; Barber-Meyer et al. 2008; Ripple & Beschta 2012a).

Dogs are the most common predator of wildlife worldwide (Ritchie et al. 2014); nevertheless, their

effects on structuring ecosystems remain largely unknown (Hughes & Macdonald 2013; Lescureux

& Linnell 2014).

Here, we aimed to understand (1) the relative top-down effects of apex predators on

mesopredators and herbivores relative to the indirect effects of humans via their land-use; and (2)

the direct and indirect effects of human presence throughout the landscape on the interactions

between apex predators, mesopredators, and herbivores. We tested specific a priori expectations

within a conceptual framework using piecewise structural equation modelling as outlined below.

Methods


Our study area covered 4900 km2 in the foothills of the Carpathian Mountains in Southern

Transylvania, Romania (Fig. 5.1). The region contains 28% forest, 24% pasture, and 37% arable

land. The remaining land cover included villages, water bodies, and permanent crops. Forests are

dominated by hornbeam (Carpinus betulus), oak (Quercus sp.), and beech (Fagus sylvatica). Pastures

occupy the hills and are grazed by sheep (dominant livestock), goats and cattle, which are guarded

by shepherds and guard dogs. Small semi-subsistence farming villages of up to several hundred

inhabitants are scattered throughout the study area (Fig. 5.1).

We surveyed wild mammal, human, and dog presence in forests using remote, heat and

motion, passive infra-red Bushnell Trophy Cam HD Max cameras between May and August 2013.

Camera locations were selected according to two considerations. First, we divided the study area

into grid cells of 5 km x 5 km, and excluded all grid cells with less than 20% forest cover (n = 120).

Chapter V

112

We placed one camera in the middle of each grid cell or in the nearest forest patch if no forest was

present there (Fig. 5.1). Cameras were rotated in 4 consecutive rounds with grid-cells randomly

allocated to each round. Second, an additional set of cameras in round 3 and 4 were placed within

24 out of 35 hunting blocks with known lowest and highest wolf densities (n = 59, for hunting

blocks see below; Fig. 5.1). Camera locations were chosen randomly but in proportion to forest

cover within each block (one camera per 5-7 km2 forest). For both designs, cameras were spaced

with a minimum distance of 1.5 km to minimize spatial autocorrelation.

Figure 5.1 Study area in Southern Transylvania, Romania, camera trap locations, and wolf densities (km2) for

each hunting block. The white areas indicate hunting blocks for which no data was available.

We used a total of 179 camera locations, with individual cameras operating between 15 and 29

days. However, only 138 locations were used for modelling because we excluded all cameras that

operated for less than 20 days (n = 28; Hamel et al. 2012) or were located in hunting blocks for

which no predator density data could be obtained (n = 12). In addition, seven cameras were stolen

and one camera recorded 26 bear presences and was removed as a statistical outlier.

To increase the chance of predator detection, we placed cameras alongside animal and

human paths, and used a lure of honey and wolf urine to attract bears and wolves, respectively.

Lures were deployed at 75% of the locations (selected randomly), while the other 25% served as

controls to assess whether wolf urine deterred herbivores. Since the presence of lures did not affect

species occurrence of predator and prey (Fig. S5.1), we did not consider lures in further analyses.


113

Variables used for modelling

We modelled the occurrence of five species in relation to top-down and bottom-up variables: bears

and dogs (apex predators), foxes (mesopredator; Vulpes vulpes), and red and roe deer (herbivores;

Cervus elaphus and Capreolus capreolus, respectively). Species occurrence was derived from data

collected by the cameras, and calculated as a cumulative abundance index or encounter rate by

summing all individuals for each species at each camera location, and corrected for camera days

during modelling (see below). Due to insufficient wolf records (two presences), wolves were not

included as a response variable.

Explanatory variables included (i) top-down variables which were represented by apex

predators (wolves, bears, dogs) and humans; and (ii) bottom-up variables which were represented

by land-use variables (forest and pasture cover). Bear and dog variables were calculated from the

camera data as the number of presences per camera day. Because of insufficient wolf records and

surprisingly little spatial variation in bear presences across camera locations, we obtained additional

information on large-scale wolf and bear densities within the 35 hunting blocks from 2010

(http://www.mmediu.ro/paduri/vanatoare.htm-accessed 01.02.2014). Although absolute densities

are most likely overestimated (Salvatori et al. 2002), this data provided a useful general indicator for

regional-scale differences in predator pressure. Notably, reported predator densities had been

largely stable between 2006 and 2010 (Supplementary Material Fig. S5.2, S5.3), suggesting that the

2010 data was likely to be indicative of bear and wolf densities, despite not being from the same

year as our camera data. The local density of humans near camera traps was calculated as the

number of presences per camera day. As an additional approximation of human pressure, we also

calculated the total number of people within the three nearest villages to each camera location. For

bears, we used either reported hunting block bear density or bear encounter rate from the camera as

an explanatory variable, and for humans, nearby village population size or human encounter rate

from the camera. The choice between these two variables was based on which provided the better

fit for a given response variable, based on Akaike information criterion, AIC (i.e. lower AIC

values). Pasture cover (range 0 - 50%; median: 13%) and forest cover (range 15 - 100%; median:

59.5%) were derived from the Corine Land Cover map (2006) within a radius of 1000 m around

camera locations. Forest cover also indicated fragmentation because it was highly correlated with

the forest edge to area ratio (ρ = -0.83).

Modelling

We used piecewise structural equation modelling (SEM) to model the importance of top-down and

bottom-up effects for the five target species. SEMs are used to analyse both direct and indirect

relationships in ecosystem processes, where a priori knowledge of relationships between

components is available (Grace 2006). In contrast to classical SEM, a piecewise approach does not

calculate global estimates for the entire network of relationships, but calculates local estimates for

Chapter V

114

each ‘node’ or response variable (Grace et al. 2012). This approach has been applied in recent

studies of trophic cascades (Pasanen-Mortensen et al. 2013; Colman et al. 2014).

First, generalized linear mixed-effect models with a Poisson error distribution were built

for each species. Depending on the species, fixed effects included human variables, natural

predators and competitors, and bottom-up variables (see Model description). All fixed variables were

scaled, and log-transformed where transformation led to a better explanation of the response

variable (based on Akaike information criterion, AIC). Random effects included hunting block,

study round, and – in cases of overdispersion (i.e. φ >1.5) – individual site identity (Elston et al.

2001). We included camera days as an offset in all models to account for differences in exposure

time for response variables.

Next, we performed model averaging across the set of models with all possible

combinations of fixed variables. We calculated averaged parameter estimates, averaged standard

errors, and the relative importance of explanatory variables (i.e. the sum of AIC weights over all

models including the explanatory variable) on the subset of models within dAIC<4 of the best

model. We also calculated marginal R2 for the full model to assess the explained variance of the

fixed factors (Nakagawa & Schielzeth 2013). Finally, composite graphs of all local estimates were

generated to visualize the relative importance of relationships between ecosystem components. All

statistical analyses were performed in R using the packages lme4 and MuMIn (R Core Team 2013).

Model description

The pathways between top-down and bottom-up variables were determined by a priori knowledge

on trophic cascade theory and included the following assumptions (Fig. 5.2). Apex predators

(brown bear, wolf, dog) were assumed to potentially limit the main mesopredator, the red fox,

through interference competition or intraguild predation (Elmhagen & Rushton 2007; Ritchie &

Johnson 2009), as well as to limit red deer and roe deer through direct predation (Peterson et al.

2003; Ripple & Beschta 2012a). Bottom-up factors were assumed to be strong for the brown bear

with an expected positive effect of forest cover and a negative effect of pasture cover (Dorresteijn

et al. 2014). We did not assume a relationship between dogs and forest cover, but expected a strong

positive link with pasture cover because dogs are commonly used for shepherding in our study

area. We assumed bottom-up factors would play a role in mesopredator and herbivore encounter

rate, but that top-down would be stronger than bottom-up effects as expected for productive

ecosystems (Ecosystem Exploitation Hyoptthesis; Oksanen & Oksanen 2000). We expected the

fox encounter rate to be negatively affected by forest cover and positively affected by pasture

cover since this species prefers fragmented and open farmland (Kurki et al. 1998; Panek &

Bresiński 2002). Herbivores were expected to be positively affected by both forest and pasture

cover (Godvik et al. 2009; Morellet et al. 2011). We assumed that there would be no intraguild

competition between wolves and bears due to different diets, but that wolves would positively

affect bear encounter rate through increasing carrion (Wilmers et al. 2003). In contrast, we assumed


115

that wolves and bears potentially limit dog encounter rate through interference competition. In

addition, we expected that roe deer would be limited by red deer through interspecific competition

(Torres et al. 2012).

Humans were expected to limit all other species, except dogs, through habitat modification

and disturbance (including instilling fear), or direct killing by hunting or poaching. Humans were

expected to indirectly limit red fox and herbivore encounter rate through their positive effects on

dogs. We did not include a link between humans and wolves because of too few camera records of

wolves. We did not attempt to explain wolf or bear densities obtained for hunting blocks, because

these were at a much larger scale than species encounter rates obtained from cameras or human

population size in nearby villages.

Figure 5.2 A priori piecewise structural equation model describing hypothesized predator-prey interactions in

a human-dominated rural landscape. Positive links are indicated by solid blue lines and negative links are

indicated by dashed red lines.

Forest cover

Cervus elaphus

Pasture cover


Capreolus capreolus

Homo sapiens

Vulpes vulpes

positive

negative

Chapter V

116

Results

Across 3042 camera days at 138 locations we obtained 2197 detections of roe deer, 388 of red

foxes, 275 of humans, 120 of dogs, 94 of red deer, 76 of bears, and 2 of wolves.

Top-down and bottom-up effects on species encounter rates

In accordance with our a priori SEM, wolves and bears had a negative effect on foxes (Fig.

5.3a). Bottom-up effects were similar in strength as top-down effects on fox encounter rates

with the strongest negative effect found for forest cover (Fig. 5.3a). Nevertheless, both top-

down and bottom-up effects were fairly weak with standard errors indicating a high degree of

uncertainty, and explained less variance as compared to all other species (Fig. 5.3a). Positive

bottom-up effects of forest cover and negative effects of pasture cover were strong

determinants of bear encounter rates (Fig. 5.3a). Within the apex predator guild, bears were

negatively related to wolves (Fig. 5.3a).

Figure 5.3 Final structural equation models showing top-down and bottom-up pathways for encounter rates

of (A) predators, and (B) herbivores. Co-efficient estimates and standard errors are presented for each

interaction and marginal R2 are given for all response variables (i.e. brown bear, domestic dog, red fox, red

deer, and roe deer). The line thickness of the arrows refers to the Akaike weights and can therefore only be

compared within a species and not between species. Positive links are indicated by solid lines and negative

links are indicated by dashed lines. The * indicates log-transformed variables

Forest cover

Cervus elaphus

Pasture cover


Capreolus capreolus

Homo sapiens

Vulpes vulpes

Forest cover

Cervus elaphus

Pasture cover


Capreolus capreolus

Homo sapiens

Vulpes vulpes

Predators Herbivores

(A)

-0.31±0.13* -0.28±0.19* -0.16±

0.10*

-0.12±0.06*

0.01

±0.

14* 0.05±

0.06*

[R 2 = 0.13] [R

2 = 0.31]

[R2 = 0.09]

[R2 = 0.24] [R

2 = 0.76]

(B)

positive negative


117

For herbivores, we also found a negative effect of wolves on red deer and of bears on roe

deer. In contrast, a positive effect was found for wolves on roe deer and bears on red deer (Fig.

5.3b). Bottom-up effects were positive (except for forest cover on roe deer) and very weak with a

larger degree of uncertainty compared to top-down effects for both red deer and roe deer (Fig.

5.3b). For example, the negative coefficient indicating the effect of wolves on red deer was 2.6

times larger than the positive coefficients of pasture cover. Roe deer appeared further limited

through competition with red deer with its coefficient similar to the coefficients of top-down

predator effects (Fig. 5.3b).

Top-down control by humans and their dogs

As predicted, humans limited all other species (Fig. 5.3a, b). While the effect was negligible for

foxes, human top-down control on bears, dogs, and herbivores was relatively strong (Fig. 5.3a, b).

For example, the coefficient describing the effect of humans on bears was 5.6 to 8.3 times larger

than the coefficients of the bottom-up effects. Human effects on red deer and roe deer were also

larger than top-down predator effects (red deer coefficients were 1.7 - 2.5 times larger; roe deer

coefficients were 1.1 - 1.6 times larger) and much larger than the very weak bottom-up effects.

Additionally, humans had significant indirect effects on species’ encounter rates due to a

strong positive effect on dogs and their subsequent flow-on effects (Fig. 5.3a). The top-down

limiting effect of dogs on red deer and roe deer was only slightly weaker than that of natural

predators but stronger than bottom-up processes (Fig. 5.3b). In contrast, dogs were positively

correlated with fox encounter rates; however, the SEM showed a negative effect of wolves on both

dogs and foxes (Fig. 5.3a). All model selection tables are available in Supplementary Material Tables

S5.1- S5.5

Discussion

Despite growing interest in using apex predators for ecological restoration (Ritchie et al. 2012), and

although agriculture covers 40% of the world’s ice-free land surface (Foley et al. 2005), few studies

have examined and quantified the ecological role of apex predators in human-dominated

landscapes. Our study addressed (1) the relative contributions of top-down limitation by apex

predators and (2) direct and indirect human bottom-up and top-down processes on mesopredator

and herbivore encounter rates in a multiple-predator, human-dominated landscape. In accordance

with trophic cascade theory, apex predators appeared to be important in structuring the ecosystem,

particularly through the suppression of herbivores. However, the extent of human direct and

indirect top-down effects at multiple trophic levels notably had a stronger effect on the ecosystem

than apex predators. Our results suggest that human factors need far greater consideration in

trophic ecology research and associated theory, because excluding humans could lead to incomplete

understanding of ecosystem structure and function.

Chapter V

118

Mesopredator limitation by apex predators and human-mediated bottom-up effects

The mesopredator release hypothesis predicts top-down control of mesopredators by apex

predators, with bottom-up factors mediating mesopredator abundance, but not regulating it

(Crooks & Soulé 1999; Elmhagen et al. 2010). We found limited evidence of suppression of foxes

by wolves and bears in our study, therefore questioning the mesopredator release hypothesis in this

human-dominated landscape. The suppressive effects of apex predators, although present, were

generally weak, and similar in strength to bottom-up control. There are large differences in body-

size and diet between foxes, omnivorous bears, and large ungulate-predating wolves, and body size

differences may be too pronounced to provide ecological benefits to bears and wolves by killing

foxes (Donadio & Buskirk 2006; Ritchie & Johnson 2009). Similarly, foxes were not suppressed by

wolves in Europe (Pasanen-Mortensen et al. 2013), but were limited by the smaller-sized dingo,

Canis lupus dingo, in Australia (Johnson & VanDerWal 2009), and lynx, Lynx lynx, in Europe

(Elmhagen et al. 2010; Pasanen-Mortensen et al. 2013). An alternative explanation is that, at

currently low apex predator densities, mesopredators may not be significantly suppressed. Indirect

effects of interference competition, such as restricting habitat use and prey availability by instilling

fear, could also have contributed to the limiting effects of apex predators but were unaccounted for

in this study (Creel & Christianson 2008; Shirley et al. 2009; Brook et al. 2012).

Human-mediated bottom-up effects were apparent through a preference of foxes for less

forested areas that are more fragmented (see also Kurki et al. 1998). Nevertheless, this effect was

also weak, probably because generalist foxes can thrive in both forested and open landscapes

(Panek & Bresiński 2002; Payne et al. 2014), and their densities increase along a natural-urban

environmental gradient (Šálek et al. 2014). Thus, our gradients in land-use may not have been

strong enough to capture fox habitat preferences, or fox distribution may be more affected by

other human-mediated bottom-up effects such as the presence of anthropogenic food sources

(Panek & Bresiński 2002). Furthermore, increased spatial heterogeneity in more complex habitats

could have mediated interference competition and dampened top-down suppression of foxes

through reduced encounter rates (McGee et al. 2006; Ritchie & Johnson 2009). More importantly,

bottom-up effects influenced foxes and bears, and most likely wolves (Jedrzejewski et al. 2004), in

opposite ways and thus, human-mediated bottom-up processes could further reduce interference

competition between foxes and apex predators through increasing forest loss and fragmentation in

Romania (Knorn et al. 2012; Griffiths et al. 2013).

Herbivore limitation by apex predators and human-mediated bottom-up effects

Despite human presence, apex predators still exerted substantial top-down limitation on

herbivores. This is consistent with the Ecosystem Exploitation Hypothesis for systems with tri-

trophic-cascades, where herbivores should be top-down limited and apex predators bottom-up

limited (Oksanen & Oksanen 2000). Top-down control of wolves and bears on red deer and roe

deer showed varying patterns. The observed negative effect on red deer and not on the roe deer


119

encounter rate by wolves can be explained by wolves’ preference for red deer over roe deer when

both species are present (Jȩdrzejewski et al. 2000). Wolf extirpation caused eruptions of deer

populations in European and American national parks (Jedrzejewski et al. 2002; Ripple & Beschta

2012b), and our results suggest that similar wolf reductions and extirpations could have led to

increased red deer populations elsewhere in Europe (Lovari et al. 2007). By contrast, roe deer

populations were only suppressed by apex predators in unproductive landscapes, and were more

affected by foraging needs and competition for food in Europe (Melis et al. 2009; Samelius et al.

2013).

In contrast, bears had a negative effect on roe deer and not on red deer. Bears are known

to predate young cervids (Berger et al. 2001; Swenson et al. 2007), and may either have predated

preferentially on fawns of roe deer over those of red deer, or the positive relationship between

bears and red deer is a likely indirect effect of the strong negative impact of humans on both

species. However, since the diet of bears in our region does not include a lot of meat (Bojarska &

Selva 2012), deer could alternatively have been suppressed by apex predators through a landscape

of fear where deer alter their behaviour in response to predation risk (Laundré et al. 2001; Creel &

Christianson 2008).

Although our results confirm theoretical predictions of weak bottom-up effects on deer

species, they contrast with a recent study where human-mediation of forage quality influenced

herbivores more strongly than top-down predator effects (Muhly et al. 2013). A lack of bottom-up

effects may be due to the present land cover composition, which features forest and pasture cover

close to the 30% threshold below which fragmentation effects become severe (Andrén 1994).

Alternatively, the resolution of the Corine Land Cover (25 ha minimum mapping unit) may have

been too coarse to pick up deer habitat preferences. Although human-mediated processes through

land-use may not determine deer encounter rates in our study area; other human-mediated bottom-

up processes such as supplemental feeding of deer may affect their populations (Schmidt & Hoi

2002). Moreover, similar to mesopredators, an increase in deforestation in the area would reduce

apex predator presence and ultimately reduce top-down control of deer.

Humans as apex predators in the system

Direct and indirect human top-down impacts were more important in shaping patterns of species

encounter rates compared to the effects of apex predators and human-mediated bottom-up effects.

Thus, our study shows that humans themselves are an apex predator in the system, indicating that

they should not be ignored in predator-prey studies (Sergio et al. 2014), particularly given the

pervasive impacts of humans across the globe (Steffen et al. 2007; Goudie 2013). Humans are

perhaps unique amongst apex predators in their ability to influence ecosystems through

simultaneously directly reducing large carnivore, mesopredator and herbivore populations and,

impacting their behaviour by creating a landscape of fear for all three trophic levels (Ordiz et al.

2013).

Chapter V

120

Direct human effects on foxes were negligible in our study. This is consistent with Baker &

Harris (2006) who showed that fox culling through hunting does not necessarily reduce fox

numbers. In contrast, direct human negative effects on deer were relatively large. Hunting of deer

could have directly reduced deer populations (see e.g. Jȩdrzejewski et al. 2000), however, the

observed pattern could also be a response to an anthropogenic landscape of fear where deer avoid

areas where hunting and other human activities are prevalent (Theuerkauf & Rouys 2008). In

addition, humans also suppressed deer through the use of livestock guard dogs, which are kept on

the pasture and thus in proximity to preferred deer foraging areas. Although the effects of dogs on

wildlife are relatively unknown, they can reduce herbivore populations through the same

mechanisms of direct predation and behaviourally induced changes as other apex predators (Ritchie

et al. 2014).

Indirectly, humans suppression of bears and possibly wolves (Jedrzejewski et al. 2004;

Llaneza et al. 2012) could lift top-down control and lead to increased herbivore populations and

possibly further mesopredator release. Herbivore and mesopredator population increases after

anthropogenic extirpations of apex predators are widely documented (Terborgh & Estes 2010). For

example, in Europe, the loss of lynx due to a combination of anthropogenic pressures

(Breitenmoser et al. 2000), caused large scale mesopredator release in Europe (Elmhagen et al.

2010), perhaps even in our study area. However, here we found that in a system where both

humans and other apex predators are present, top-down control by humans and predators and

indirect release due to human suppression of apex predators act simultaneously, particularly on

herbivore populations. Further studies that disentangle the effects of humans and predators on

lower trophic levels will be key to advancing our understanding of the drivers and dynamics of

ecosystems, trophic cascade theory, and ultimately how these affect biodiversity conservation. This

is especially important since humans may not replicate the exact nature of indirect effects caused by

other apex predators, highlighting that the ecological roles of apex predators are not always

interchangeable (Ordiz et al. 2013). This may explain in part why humans are often unsuccessful at

preventing or reversing negative impacts such as overgrazing, reduced vegetation recruitment, and

biodiversity loss caused by altered predator-herbivore-plant trophic cascades or mesopredator

release (e.g. Terborgh et al. 2001; Estes et al. 2011; Kuijper 2011; Ripple et al. 2013).

Conclusions

To date, ecological theory on trophic cascades has not explicitly included human effects, despite

humanity’s pervasive impacts on the globe (Sanderson et al. 2002). Our study adds to a growing

recognition that humans play vital roles in influencing ecosystems through mediating and altering

trophic cascades as well as through direct landscape modification. Apex predators maintained their

ecological role by suppressing lower trophic levels in a human-dominated landscape, but both

direct and indirect anthropogenic top-down effects dominated over natural processes. Improving

our understanding of human impacts on trophic cascades in human-dominated landscapes is


121

especially important because apex predators are declining rapidly in much of the world, but just as

importantly they are also being encouraged to recover and are reintroduced to other areas. There

are ample possibilities for restoring ecosystems through rewilding efforts or carnivore

reintroduction programs, but especially in this context, it is important to anticipate the implications

of simultaneous effects of humans and apex predators on multiple trophic levels. Our results and

conclusions are presented with caution, as we acknowledge the limitations of correlative studies

and the need to test conceptual understanding via experiments, and in doing so establish

mechanistic causation underpinning the patterns observed. Given the extent and speed of global

anthropogenic environmental change, elucidating how humans directly and indirectly alter bottom-

up and top-down processes should receive increased consideration by future studies.

Acknowledgements

We thank Martin Bubner, Frank Dietrich, Lunja Ernst, Cathy Klein, Jenny Long, Frans Meeuwsen,

Arpad Szapanyos, Joris Tinnemans, Nathanaël Vetter for tremendous field efforts and data entry.

The project was funded by a Sofja Kovalevskaja Award by the Alexander von Humboldt

Foundation to JF and the Einstein Foundation Berlin to TK and LK.

Chapter V

122

Supplementary Material for Chapter V

Figure S5.1 Boxplots of species encounter rates per camera day at camera locations with and without lure.

The boxplots show that the presence of the lure did not affect species encounter rates appreciably and was

therefore not included as a random factor in the models.

Figure S5.2 5 year comparison of reported wolf densities per hunting block. Wolf densities per hunting

block (i.e. number of reported wolves divided by hunting block size; http://www.mmediu.ro/paduri/

vanatoare.htm - accessed 01.02.2014) were ranked within each year (high rank = low density, low rank = high

density). Ranks of hunting blocks for 2006-2009 were then plotted against 2010 ranks. Overall, the ranks of

hunting block wolf densities were similar throughout years.


123

Figure S5.3 5 year comparison of reported bear densities per hunting block. Bear densities per hunting block

(i.e. number of reported bears divided by hunting block size; http://www.mmediu.ro/paduri/vanatoare.htm

- accessed 01.02.2014) were ranked within each year (high rank = low density, low rank = high density).

Ranks of hunting blocks for 2006-2009 were then plotted against 2010 ranks. Overall, the ranks of hunting

block bear densities were similar throughout years.

Table S5.1 Model selection table for all candidate models for brown bear. Average estimates are given for

each variable. Log stands for log-transformation of the variable. Human (B) indicates human population size

in the nearest three villages.

Rank Forest (log) Pasture Human (B) Wolf (log) df

Log likelihood AICc

delta AICc weight

1 + + + + 7 -131.09 277.00 0.00 0.42

2

+ + + 6 -132.66 278.00 0.92 0.26

3 + + +

6 -133.42 279.50 2.42 0.12

4

+ +

5 -134.81 280.10 3.02 0.09

5 +

+ + 6 -133.79 280.20 3.17 0.09

6 +

+

5 -136.89 284.20 7.19 0.01

7

+ + 5 -137.10 284.70 7.61 0.01

8

+

4 -140.05 288.40 11.36 0.00

9 + +

+ 6 -138.58 289.80 12.75 0.00

10 + +

5 -139.74 289.90 12.88 0.00

11 +

+ 5 -140.82 292.10 15.05 0.00

12

+

4 -142.44 293.20 16.13 0.00

13 +

4 -142.52 293.30 16.29 0.00

14

+

+ 5 -141.61 293.70 16.63 0.00

15

+ 4 -145.40 299.10 22.04 0.00

16

3 -146.84 299.90 22.81 0.00

Coefficient 0.25 -0.38 -2.07 -0.30

http://www.mmediu.ro/paduri/vanatoare.htm%20-%20accessed%2001.02.2014

http://www.mmediu.ro/paduri/vanatoare.htm%20-%20accessed%2001.02.2014

Chapter V

124

Table S5.2 Model selection table for all candidate models for domestic dog. Average estimates are given for

each variable. Log stands for log-transformation of the variable. Bear (B) indicates bear densities from the

hunting blocks. Human (L) indicates number of humans from the camera traps.

Rank Bear (log, B)

Human (log, L) Pasture Wolf (log) df

Log likelihood AICc

delta AICc weight

1

+ +

5 -190.23 390.90 0.00 0.37

2

+ + + 6 -189.63 391.90 1.00 0.22

3 + + +

6 -189.83 392.30 1.39 0.19

4 + + + + 7 -188.77 392.40 1.50 0.18

5

+

4 -194.16 396.60 5.71 0.02

6 + +

5 -193.87 398.20 7.28 0.01

7

+

+ 5 -193.94 398.30 7.43 0.01

8 + +

+ 6 -193.41 399.50 8.54 0.01

9 +

+ + 6 -209.17 431.00 40.07 0.00

10

+

4 -211.40 431.10 40.20 0.00

11

+ + 5 -210.49 431.40 40.53 0.00

12 +

+

5 -210.78 432.00 41.11 0.00

13

3 -215.07 436.30 45.42 0.00

14 +

+ 5 -213.37 437.20 46.28 0.00

15

+ 4 -214.48 437.30 46.36 0.00

16 +

4 -214.50 437.30 46.39 0.00

Coefficient 0.25 0.63 0.37 -0.27

Table S5.3 Model selection table for all candidate models for red fox. Average estimates are given for each

variable. Log stands for log-transformation of the variable. Bear (B) indicates bear densities from the hunting

blocks. Human (L) indicates number of humans from the camera traps.

Rank Bear (log, B) Dog Forest

Human (log, L)

Pasture (log)

Wolf (log) df

Log likelihood AICc

delta AICc weight

1

+

+ 6 -295.73 604.10 0.00 0.06

2

+ +

+ 7 -294.75 604.40 0.27 0.05

3

+

5 -297.00 604.50 0.36 0.05

4 +

+

6 -295.95 604.50 0.44 0.05

5 + + +

7 -294.84 604.50 0.45 0.05

6

+ +

6 -296.02 604.70 0.59 0.05

7 +

+

+ 7 -295.22 605.30 1.21 0.03

8

+

+ 6 -296.37 605.40 1.29 0.03

9 + + +

+ 8 -294.16 605.40 1.33 0.03

10

+

+ + 7 -295.48 605.80 1.72 0.03

11

+

+ + 7 -295.56 606.00 1.88 0.02

12

+ +

+ + 8 -294.44 606.00 1.90 0.02

13 + +

6 -296.81 606.30 2.16 0.02

14

+ +

+ 7 -295.71 606.30 2.19 0.02

15

+ + +

+ 8 -294.61 606.30 2.24 0.02

16 + +

+ 7 -295.78 606.40 2.32 0.02

17 + + +

+

8 -294.67 606.50 2.35 0.02

18

+ 5 -298.02 606.50 2.39 0.02

19

+

+

6 -296.94 606.50 2.43 0.02

20

+ +

+

7 -295.87 606.60 2.50 0.02

21

+ +

6 -296.99 606.60 2.53 0.02

22 +

+

+

7 -295.88 606.60 2.53 0.02

23

+ + +

7 -295.91 606.70 2.58 0.02

24 + + + +

8 -294.79 606.70 2.60 0.02

25

+

5 -298.14 606.70 2.63 0.02

26 +

+ +

7 -295.95 606.80 2.66 0.02

27 + +

+ + 8 -294.91 606.90 2.84 0.02

28 + +

+

7 -296.10 607.10 2.97 0.01

29 + + +

+ + 9 -293.87 607.10 3.04 0.01

30

+ + 6 -297.28 607.20 3.10 0.01

31 +

+

+ + 8 -295.08 607.30 3.17 0.01


125

32

+

+

6 -297.45 607.50 3.43 0.01

33 + + + +

+ 9 -294.07 607.60 3.45 0.01

34 +

+ +

+ 8 -295.22 607.60 3.46 0.01

35

+

+

+ 7 -296.37 607.60 3.50 0.01

36 +

+ 6 -297.54 607.70 3.62 0.01

37 +

5 -298.72 607.90 3.80 0.01

38

+ + + + + 9 -294.28 608.00 3.87 0.01

39

+

+ + + 8 -295.44 608.00 3.91 0.01

40

4 -299.90 608.10 3.99 0.01

41

+ + + + 8 -295.54 608.20 4.10 0.01

42 + +

+

7 -296.80 608.50 4.35 0.01

43

+

+ 6 -297.93 608.50 4.40 0.01

44 +

+ + 7 -296.83 608.50 4.42 0.01

45

+ + + +

8 -295.75 608.60 4.51 0.01

46 + + + + +

9 -294.61 608.60 4.53 0.01

47 + +

+

+ 8 -295.78 608.70 4.57 0.01

48

+ + +

7 -296.93 608.70 4.63 0.01

49 +

+ + +

8 -295.88 608.90 4.78 0.01

50

+

+

6 -298.13 608.90 4.81 0.01

51 +

+

6 -298.17 609.00 4.87 0.01

52

+

5 -299.35 609.20 5.05 0.01

53 + +

+ + + 9 -294.90 609.20 5.11 0.01

54 + + + + + + 10 -293.77 609.30 5.18 0.01

55

+ + + 7 -297.23 609.30 5.22 0.00

56 + +

+ +

8 -296.10 609.30 5.22 0.00

57 +

+ + + + 9 -295.07 609.60 5.46 0.00

58 +

+

6 -298.46 609.60 5.46 0.00

59 +

+

+ 7 -297.39 609.60 5.54 0.00

60

+

+ +

7 -297.44 609.70 5.64 0.00

61

+

5 -299.73 609.90 5.82 0.00

62 +

+ + + 8 -296.73 610.60 6.48 0.00

63 +

+ +

7 -297.97 610.80 6.70 0.00

64

+ +

6 -299.23 611.10 7.01 0.00

Coefficient -0.13 0.15 -0.22 -0.03 0.09 -0.16

Table S5.4 Model selection table for all candidate models for red deer. Average estimates are given for each


blocks. Human (B) indicates human population size in the nearest three villages.

Rank Bear (log, B)

Dog (log)

Forest )log)

Pasture (log)

Human (log, B) Wolf df

Log likelihood AICc

delta AICc weight

1 + +

+ + 7 -160.84 336.50 0.00 0.18

2 + +

+ + + 8 -160.14 337.40 0.86 0.11

3

+

+ + 6 -162.62 337.90 1.34 0.09

4

+

+

5 -163.79 338.00 1.49 0.08

5 + +

+

6 -162.96 338.60 2.02 0.06

6 + + +

+ + 8 -160.84 338.80 2.25 0.06

7

+

+ + + 7 -162.03 338.90 2.39 0.05

8

+

+ +

6 -163.24 339.10 2.58 0.05

9 + +

+ +

7 -162.35 339.60 3.02 0.04

10 + + + + + + 9 -160.10 339.60 3.07 0.04

11

+ +

+ + 7 -162.62 340.10 3.56 0.03

12

+ +

+

6 -163.78 340.20 3.67 0.03

13 +

+ + 6 -164.02 340.70 4.14 0.02

14 + + +

+

7 -162.96 340.80 4.23 0.02

15

+ + + + + 8 -162.02 341.20 4.61 0.02

16

+ + + +

7 -163.24 341.30 4.79 0.02

17

+ + 5 -165.45 341.40 4.81 0.02

18

+

4 -166.53 341.40 4.83 0.02

19 + + + + +

8 -162.35 341.80 5.27 0.01

20 +

+

5 -165.89 342.20 5.68 0.01

21 +

+ + + 7 -163.92 342.70 6.16 0.01

22 +

+

+ + 7 -163.95 342.80 6.22 0.01

23

+ + + 6 -165.37 343.40 6.83 0.01

24

+ +

5 -166.47 343.40 6.85 0.01

25

+

+ + 6 -165.40 343.40 6.91 0.01

26

+

+

5 -166.51 343.50 6.94 0.01

27 +

+ +

6 -165.81 344.30 7.73 0.00

28 +

+

+

6 -165.86 344.40 7.83 0.00

29 +

+ + + + 8 -163.79 344.70 8.16 0.00

Chapter V

126

30

+ + + + 7 -165.28 345.40 8.88 0.00

31

+ + +

6 -166.42 345.50 8.94 0.00

32 +

+ + +

7 -165.76 346.40 9.84 0.00

33 + +

+

+ 7 -169.15 353.20 16.62 0.00

34 + + + +

+ 8 -168.73 354.60 18.03 0.00

35 + +

+

6 -171.56 355.80 19.22 0.00

36

+

+

5 -173.11 356.70 20.14 0.00

37

+

+

+ 6 -172.04 356.70 20.17 0.00

38 + + + +

7 -171.26 357.40 20.85 0.00

39 + +

+ 6 -172.56 357.80 21.21 0.00

40

+ + +

+ 7 -171.63 358.10 21.58 0.00

41

+ + +

6 -172.79 358.20 21.68 0.00

42 + +

5 -174.56 359.60 23.04 0.00

43 + + +

+ 7 -172.37 359.60 23.06 0.00

44

+

4 -176.01 360.30 23.78 0.00

45

+

+ 5 -175.15 360.80 24.22 0.00

46 + + +

6 -174.44 361.50 24.99 0.00

47

+ +

5 -175.87 362.20 25.66 0.00

48

+ +

+ 6 -174.97 362.60 26.04 0.00

49 +

+ +

+ 7 -174.82 364.50 27.97 0.00

50 +

+

+ 6 -176.02 364.70 28.14 0.00

51 +

+ 5 -177.52 365.50 28.96 0.00

52 +

+

+ 6 -176.83 366.30 29.76 0.00

53 +

+

5 -178.02 366.50 29.96 0.00

54

+

4 -179.18 366.70 30.12 0.00

55 +

+ +

6 -177.04 366.70 30.18 0.00

56

+ +

+ 6 -177.04 366.70 30.19 0.00

57

+ +

5 -178.16 366.80 30.23 0.00

58 +

4 -179.27 366.80 30.30 0.00

59

+

+ 5 -178.22 366.90 30.35 0.00

60

3 -180.43 367.00 30.50 0.00

61

+ 4 -179.64 367.60 31.04 0.00

62 +

+

5 -178.71 367.90 31.32 0.00

63

+

4 -179.84 368.00 31.44 0.00

64

+

+ 5 -178.95 368.40 31.82 0.00

Coefficient 0.44 -0.32 0.01 0.18 -0.84 -0.48

Table S5.5 Model selection table for all candidate models for roe deer. Average estimates are given for each


blocks. Human (B) indicates human population size in the nearest three villages.

Rank Bear (log, B)

Dog (log) Forest

Pasture (log)

Human (B)

Red deer (log) Wolf df logLik AICc delta weight

1

+ +

6 -494.92 1002.50 0.00 0.05

2 +

+ + + 8 -492.70 1002.50 0.02 0.05

3 +

+ +

7 -493.95 1002.80 0.26 0.04

4

+

+ +

7 -494.01 1002.90 0.39 0.04

5 +

+

+ 7 -494.28 1003.40 0.94 0.03

6 + +

+ +

8 -493.27 1003.60 1.16 0.03

7 + +

+ + + 9 -492.19 1003.80 1.30 0.03

8

+ + + 7 -494.52 1003.90 1.40 0.02

9

+

+ + +

8 -493.59 1004.30 1.80 0.02

10

+ + +

7 -494.73 1004.30 1.82 0.02

11

+

+ + + 8 -493.65 1004.40 1.93 0.02

12 +

+ + + + 9 -492.54 1004.50 1.99 0.02

13 +

+ + +

8 -493.74 1004.60 2.10 0.02

14 +

+

+ + + 9 -492.62 1004.60 2.15 0.02

15

+

+ +

7 -494.92 1004.70 2.20 0.02

16

+

+

6 -496.05 1004.70 2.26 0.02

17

+ +

+ +

8 -493.88 1004.90 2.39 0.01

18 +

+

6 -496.14 1004.90 2.43 0.01

19 +

+ 6 -496.17 1005.00 2.49 0.01

20 +

+

+ +

8 -493.94 1005.00 2.50 0.01

21 +

+ + 7 -495.08 1005.00 2.53 0.01

22 + +

+

+ 8 -493.96 1005.00 2.54 0.01

23 + +

+ + +

9 -492.87 1005.10 2.65 0.01

24 +

+

+

+ 8 -494.10 1005.30 2.82 0.01

25

+

5 -497.43 1005.30 2.82 0.01

26

+

+

+

7 -495.23 1005.30 2.83 0.01


127

27

+

5 -497.45 1005.40 2.86 0.01

28 +

+ +

+ 8 -494.15 1005.40 2.92 0.01

29 + +

+ + + + 10 -491.89 1005.50 3.02 0.01

30 + +

+ + 8 -494.20 1005.50 3.03 0.01

31 + + +

+ + + 10 -491.96 1005.70 3.16 0.01

32 + + +

+ +

9 -493.16 1005.70 3.24 0.01

33 + +

+

7 -495.44 1005.70 3.24 0.01

34 +

+

6 -496.58 1005.80 3.30 0.01

35

+ + + + 8 -494.35 1005.80 3.32 0.01

36 + +

+ 7 -495.54 1005.90 3.45 0.01

37

+

+ + + + 9 -493.28 1006.00 3.47 0.01

38

+

+ + 7 -495.56 1006.00 3.49 0.01

39

+

+ + + 8 -494.48 1006.10 3.59 0.01

40

+

+ 6 -496.74 1006.10 3.63 0.01

41

+

+

6 -496.76 1006.20 3.67 0.01

42 + +

+

7 -495.67 1006.20 3.72 0.01

43

+ +

+ + + 9 -493.44 1006.30 3.80 0.01

44 + +

+

+

8 -494.63 1006.40 3.89 0.01

45

+ + 6 -496.87 1006.40 3.89 0.01

46

+ +

+

7 -495.79 1006.40 3.94 0.01

47 +

+

+ + 8 -494.68 1006.50 3.98 0.01

48

+ + + + +

9 -493.54 1006.50 4.00 0.01

49 + +

+

+ + 9 -493.55 1006.50 4.02 0.01

50 +

+

+ 7 -495.83 1006.50 4.03 0.01

51

+ + + +

8 -494.73 1006.60 4.08 0.01

52

+

+

6 -496.97 1006.60 4.09 0.01

53 + + +

+

+ 9 -493.59 1006.60 4.10 0.01

54 +

+

+ 7 -495.91 1006.70 4.19 0.01

55 +

+ + + + + 10 -492.51 1006.70 4.26 0.01

56

+

+

+ + 8 -494.82 1006.80 4.27 0.01

57 +

+ +

7 -495.96 1006.80 4.30 0.01

58 +

5 -498.17 1006.80 4.31 0.01

59 + +

+ +

+ 9 -493.72 1006.80 4.36 0.01

60 +

+ + + +

9 -493.73 1006.90 4.39 0.01

61

4 -499.32 1006.90 4.44 0.01

62 +

+

+ + 8 -494.92 1007.00 4.46 0.01

63

+

5 -498.25 1007.00 4.46 0.01

64 + + +

+ + 9 -493.78 1007.00 4.48 0.01

65 +

+

+

7 -496.08 1007.00 4.53 0.01

66 +

+

+

7 -496.09 1007.00 4.56 0.01

67

+

+

+ 7 -496.11 1007.10 4.58 0.01

68 + + +

+ 8 -494.99 1007.10 4.61 0.01

69 + +

+

+ 8 -495.02 1007.10 4.66 0.01

70

+ +

6 -497.27 1007.20 4.70 0.01

71 + +

6 -497.33 1007.30 4.82 0.00

72

+ + +

+

8 -495.13 1007.40 4.89 0.00

73

+

+

6 -497.38 1007.40 4.90 0.00

74 + + + + + +

10 -492.83 1007.40 4.91 0.00

75

+ 5 -498.50 1007.40 4.96 0.00

76

+

+

6 -497.41 1007.50 4.97 0.00

77

+ +

+ + 8 -495.17 1007.50 4.97 0.00

78 +

+ + +

+ 9 -494.03 1007.50 4.97 0.00

79 + + +

+

8 -495.20 1007.50 5.02 0.00

80 + + + + + + + 11 -491.76 1007.60 5.12 0.00

81

+

+ 6 -497.49 1007.60 5.14 0.00

82

+

+ +

7 -496.43 1007.70 5.23 0.00

83

+

+ + 7 -496.45 1007.80 5.28 0.00

84 + +

+ +

8 -495.36 1007.80 5.34 0.00

85

+

+

6 -497.60 1007.80 5.36 0.00

86

+ +

+

7 -496.54 1007.90 5.44 0.00

87 +

+

+

7 -496.54 1007.90 5.45 0.00

88

+ + + + + + 10 -493.17 1008.10 5.59 0.00

89

+ +

+ 7 -496.61 1008.10 5.60 0.00

90

+ + + + + 9 -494.34 1008.10 5.60 0.00

91

+

+

+ 7 -496.62 1008.10 5.61 0.00

92 + + +

+

8 -495.50 1008.10 5.62 0.00

93 +

+

6 -497.76 1008.20 5.67 0.00

94 + +

+

7 -496.69 1008.20 5.74 0.00

95

+

5 -498.94 1008.30 5.84 0.00

96

+

+ + 7 -496.77 1008.40 5.91 0.00

Chapter V

128

97 + + + +

+ + 10 -493.34 1008.40 5.92 0.00

98

+ +

6 -497.89 1008.40 5.94 0.00

99 +

+ +

+ 8 -495.70 1008.50 6.02 0.00

100 + + + +

+

9 -494.55 1008.50 6.02 0.00

101

+ +

+

+ 8 -495.76 1008.60 6.14 0.00

102 +

+ +

+ + 9 -494.62 1008.70 6.16 0.00

103 + + + + +

+ 10 -493.46 1008.70 6.17 0.00

104

+ + +

+ + 9 -494.65 1008.70 6.21 0.00

105

+

+

+ 7 -496.93 1008.70 6.22 0.00

106

+

+ +

+ 8 -495.82 1008.80 6.27 0.00

107 + + + +

+ 9 -494.68 1008.80 6.28 0.00

108

+ +

+

7 -496.97 1008.80 6.31 0.00

109 +

+

6 -498.09 1008.80 6.32 0.00

110

+ +

+ 7 -496.98 1008.80 6.33 0.00

111

+

5 -499.22 1008.90 6.41 0.00

112 + + +

7 -497.02 1008.90 6.42 0.00

113

+

+ 6 -498.17 1009.00 6.50 0.00

114 +

+ + +

8 -495.95 1009.00 6.53 0.00

115

+

+ 6 -498.31 1009.30 6.77 0.00

116 +

+ +

+

8 -496.08 1009.30 6.78 0.00

117

+ + +

7 -497.26 1009.40 6.89 0.00

118

+ + +

7 -497.43 1009.70 7.23 0.00

119

+ + + +

8 -496.30 1009.70 7.23 0.00

120 + + + + +

9 -495.26 1009.90 7.44 0.00

121

+ +

+ + 8 -496.44 1010.00 7.50 0.00

122 + + + +

8 -496.54 1010.20 7.71 0.00

123

+ + +

+ 8 -496.54 1010.20 7.72 0.00

124 +

+ +

7 -497.75 1010.40 7.86 0.00

125

+ + +

+ 8 -496.62 1010.40 7.87 0.00

126

+ +

6 -498.92 1010.50 7.99 0.00

127

+ + + +

+ 9 -495.59 1010.60 8.10 0.00

128

+ +

+ 7 -498.09 1011.00 8.55 0.00

Coefficient -0.12 -0.08 -0.02 0.05 -0.13 -0.12 0.11

129

Section B: Traditional

wood pastures

This Section includes Chapters VI-VIII and focuses specifically on the conservation value of

traditional wood pastures to gain insights on the role of particular traditional land-use elements for

biodiversity. Wood pastures are one of the oldest land-use types in Europe. They were established

through ancient silvo-pastoral practices and often consist of open grasslands with scattered old

trees. Chapter VI explores the characteristics, management, and status of wood pastures, but also

the current threats to wood pastures. The role of wood pastures for biodiversity is assessed in more

detail in Chapters VII and VIII: Chapter VII examines the habitat value of woodpeckers for an

assemblage of six woodpecker species, Chapter VIII aims to assess the extent of wood pastures use

by the brown bear.

130

131

Chapter VI

132

133

Chapter VI

Wood-pastures in a traditional rural region of Eastern Europe:

Characteristics, management, and status

Tibor Hartel, Ine Dorresteijn, Catherine Klein, Orsolya Máthé, Cosmin I. Moga, Kinga

Öllerer, Marlene Roellig, Henrik von Wehrden, and Joern Fischer

Biological Conservation (2013), 166:267-275

‘Someone's sitting in the shade today because someone planted a tree a long time ago.’ – Warren Buffett

‘Many a genius has been slow of growth. Oaks that flourish for a thousand years do not spring up into

beauty like a reed.’ – George H. Lewis

Chapter VI

134

Wood pasture characteristics, management and status

135

Abstract

Wood-pastures are among the oldest land-use types in Europe and have high ecological and

cultural importance. They are under rapid decline all over Europe because of changes in land-use,

tree cutting, and lack of regeneration. In this study we characterized the structure, condition and

threats of wood-pastures in a traditional rural region in Romania. Forty-two wood-pastures were

surveyed, as well as 15 forest sites for comparison. All wood-pasture sites were described via four

groups of variables: condition, management, site, and landscape context. Forest sites were

dominated by Hornbeam (Carpinus betulus) and Beech (Fagus sylvatica), whereas wood-pastures were

dominated by Oak (Quercus sp.) and various species of fruit trees. Most wood-pastures contained

trees classified as ‘ancient’ but no such trees were found in forests. The proportion of dead trees

was positively related to forest cover within 300 m around the wood-pasture. Models that included

management, site and landscape-related variables best explained the prelevance of Oak, Beech,

Hornbeam and Pear trees in wood-pastures. Large oaks and hornbeams were more likely to be

dead or affected by uncontrolled pasture burning than small oaks and other tree species. Our

results show that ancient wood-pastures are common in this rural region, and they may be more

common in Eastern Europe than previously thought. There is an urgent need for research, legal

recognition and conservation management of wood-pastures as distinct landscape elements for

their cultural, ecological and agricultural importance.

Chapter VI

136

Introduction

Wood-pastures represent an important part of European cultural–natural heritage (Bergmeier et al.

2010), and are one of the oldest land-use types in Europe, being known since the Neolithic (Luick

2008). Although the concept of wood-pastures is broad (Spencer & Kirby 1992; Caledonian

Partnership 2003; Goldberg et al. 2007), it characteristically refers to environments that are defined

by trees scattered through an open area, generally grassland. Appropriate livestock grazing regimes

applied through centuries have been crucial for the formation of wood-pastures and will be

important for their further persistence (Quelch 2002).

Ancient wood-pastures bring together several important components that make them

attractive for ecologists and conservationists. First, wood-pastures contain scattered trees. The age

of these trees can reach centuries; such trees are sometimes referred to as ‘veteran’ or ‘ancient’ trees

(Read 2000; Quelch 2002). Old, scattered trees provide a broad range of habitat features such as

dead branches or hollows (Gibbons & Lindenmayer 2003). For this reason, old trees represent

local ‘biodiversity hotspots’ in ecosystems around the world (Fischer et al. 2010; Lindenmayer et al.

2014). Moreover, scattered trees (regardless of their age) significantly influence microclimatic

conditions and soil humidity, and consequently vegetation structure (Manning et al. 2006) and may

help to facilitate adaptation to anthropogenic climate change in the future (Manning et al. 2009).

Second, the open habitat throughout which trees are scattered is managed mostly as pasture

(Quelch 2002; Mountford & Peterken 2003; Bergmeier et al. 2010). Traditional pasture

management has typically been low in intensity, thus supporting a rich flora and fauna, including

many species of conservation interest (Rosenthal et al. 2012). Low intensity grazing, together with

scattered, often old trees, makes many wood-pastures regional hotspots of biodiversity (Bugalho et

al. 2011).

Wood-pastures have received increasing scientific attention throughout Europe in recent

years. Studies on the biodiversity of wood-pastures and the ecological value of old trees have been

conducted in the Czech Republic (Vojta & Drhovská 2012; Horák & Rébl 2013), Portugal

(Gonçalves et al. 2012), Romania (Moga et al. 2009; Dorresteijn et al. 2013), and Sweden (Paltto et

al. 2011; Widerberg et al. 2012). Vegetation dynamics and landscape change related to management

regimes are available, for example, from the Swiss Jura Mountains (Buttler et al. 2008), the Italian

Alps (Garbarino et al. 2011), Belgium (Van Uytvanck et al. 2008), the Netherlands (Smit & Ruifrok

2011; Smit & Verwijmeren 2011), Spain (Plieninger & Schaar 2008) and Sweden (Brunet et al.,

2011). Studies exploring the recruitment of trees in wood-pastures are available from Spain

(Plieninger 2007); and research about the vegetation structure and conservation status of wood-

pastures is available from Romania (Öllerer 2012, 2013), Turkey (Uğurlu et al. 2012), and Greece

(Chaideftou et al. 2011). Finally, the crucial importance of low intensity human use for the

maintenance of biodiversity and ecosystem services in Mediterranean wood-pastures and their

provisioning ecosystem services was reported by Bugalho et al. (2011).


137

Existing studies highlight that wood-pastures have been undergoing major changes in the

past few decades. These changes threaten the existence of wood-pastures and are mostly driven by

changing land-use (e.g. land abandonment and changing farming practices), policies, changing

attitudes toward old trees, and lack of tree regeneration (reviewed in Bergmeier et al. 2010).

While until recently Britain was considered as one of the main locations in Europe for

large (veteran) trees in wood-pastures (Rackham 1998; Mountford & Peterken 2003), information

from Central and Eastern European (CEE) countries about ancient wood-pastures is scarce. Some

national level evaluations exist for example in Hungary (e.g Haraszthy et al. 1997), suggesting that

wood-pastures are among the most threatened ecosystems in this country.

Here, we present research on wood-pastures in a traditional rural region of Central

Romania. Our study had three aims: (i) to compare the structure of tree communities and tree sizes

between forests and wood-pastures, (ii) to describe the main characteristics and current

management of wood-pastures and (iii) to model wood-pasture condition using a number of site,

landscape and management related variables. Although our study has a regional focus and is partly

descriptive in nature, we discuss our findings broadly in the context of international wood-pasture

conservation. Drawing on our findings, we argue that some Eastern European wood-pastures have

particularly high, but largely unrecognised, conservation values.

Methods

Study area

The study was conducted in Southern Transylvania, Romania and covered ca 3600 km2, of which

ca 860 km2 were covered by Natura 2000 regulations (Site of Community Importance, hereafter

SCI) (Fig. 6.1). The region is dominated by traditional land-use practices and has low levels of

infrastructure development. The most important land cover types based on CORINE land cover

classes (see Table 6.1 for reference) are forest (ca. 30%), pasture (ca. 26%), heterogeneous

agricultural areas (including agro-forestry areas; ca. 15%) and arable fields (ca. 14%). The urban

area cover is low (ca. 3%). Other minor land covers include wetlands and vineyards. The climate in

the region is continental and moderate. Annual temperature averages 8.2 ºC, with an average

temperature of -4.3 ºC in January and 18.6 ºC in July. The yearly mean amount of precipitation is

between 650 and 700 mm (Hartel and Moga, 2010).

Chapter VI

138

Figure 6.1 Map of the study area. All survey sites, including wood-pastures, ancient wood-pastures, and

forests are shown. The Natura 2000 site (SCI) is delineated (see study area in Central Romania description).

Field methods and variables

Data were collected in 2012. Forty-two wood-pastures and 15 forest sites were studied (see

Supporting Information 6.1 and 6.2). We sampled a larger number of wood-pastures than forests

because forests are relatively homogenous, whereas wood-pastures differ substantially in structural

elements and adjacent forest cover. Furthermore, we were especially interested in the description of

wood-pastures and thus chose more sites to comprehensively cover existing gradients within wood-

pastures.

We used five groups of variables to characterize wood-pastures: condition variables (tree

density per ha, number of scattered trees per ha, proportion of dead trees per, tee size, scrub cover

and woody vegetation cover), composition variables (prevalence of oak, hornbeam, beech and

pear), site (area, elevation and ruggedness) and landscape (forest cover and distance to nearest

village) related variables and management (evidence for scrub cleaning related to Agency for

Payments and Intervention in Agriculture (hereafter APIA), livestock and burning) related

variables. Descriptions of these variables and their units of measurement and sources (for data

other than field data) are summarized in Table 6.1. The tree diameter at breast height (DBH) was

calculated from the circumference, which we measured with a tape for standing trees. Tree

measurements were made in February–March (2012) in the following way: (1) all trees in wood-

pastures were measured within a radius of 80 m around a central survey point (i.e. within a 2 ha

site). In forest sites, 50 trees were selected randomly in a spiral from the centre of the site to the

edge of the 2 ha, to obtain a representative sample of trees. (2) Additional trees were measured

within four strip transects of ca. 10 m width between 80 m and 300 m from the central survey

point in the four cardinal directions (N, S, E, W). In the case that three or fewer trees were found

within the wood-pasture transects we measured up to five trees close to the transect. The resulting

data provided indications of the diameter distribution within 2 ha around a central point and in the

immediate surroundings. We used the DBH of trees as a proxy for their age categorization and

conservation value (i.e. ‘truly ancient’, ‘ancient’, ‘of conservation value’, and ‘potentially interesting’,


139

following Read (2000) and Farm Environment Plan Guide (2006)). We also recorded if the

measured trees were burned (i.e. the tree showed signs of fire but was alive), dead (for standing

dead trees), healthy (no visible injury on the trunk of the tree) and injured (when the trunk was

injured by cutting – coppicing and pollarding were not considered as injuries).

Tree density within 2 ha was assessed in wood-pastures as the count of all trees within 80

m of the central survey point. All dead trees from the two hectare sites were counted, both in

wood-pastures and forests. Trees were identified to genus level. According to a previous study

(Hartel & Moga 2010), the vast majority of the Oaks in wood-pastures in this region belong to the

species Quercus robur (90% out of 339 Oaks measured), or to Q. petraea and hybrids between the two

species. Due to the similarities of the ecology and habit of these two oaks, we believe that

considering them together was reasonable and facilitated meaningful comparison with the other

dominant tree genera.

Scrub cover was assessed for the entire wood-pasture using 400 m long and 6 m wide

transects which were placed subjectively so that they covered all representative locations of the

wood-pasture. This assessment was made in the period of May-July. The number of transects in

each wood-pasture was chosen according to the size of the wood-pasture: two transects were used

in wood-pastures with an area of up to 30 ha, three transects in those measuring 30–80 ha, four

transects in those of 80–130 ha area, five transects in those of 130–180 ha area, and six in those

measuring more than 180 ha. On average there were 3.7 transects per site. In each transect, the

percent of scrub cover was assessed visually every 100 m. Scrub cover values were averaged for the

entire wood-pasture to obtain a single representative estimate.

Furthermore we recorded the presence/absence of livestock based on direct observation

of the animals and/or their faeces. Pasture burning was recorded in March–April period (when this

activity usually takes place as management intervention to remove excessive biomass from

pastures), and the presence of scrub removal was recorded in March–July (Table 6.1). We further

recorded if we observed tree cutting activities in wood-pastures in 2012.

Chapter VI

140

Table 6.1 The description of the environmental variables used to characterize wood-pastures from Southern

Transylvania and to model wood-pasture condition and composition. Variables highlighted in italics are those

that were used in statistical models as explanatory or response variables (see section on Analysis).

Variable name Description

(a) Condition variables Tree density per ha Calculated from the overall number of standing trees (dead and alive)

counted in the 2 ha sites. Number of scattered trees The overall number of scattered trees in the entire wood-pasture based

on counts of trees using Google Earth satellite images. Proportion of dead trees (2 ha) The percent of dead trees (standing or fallen) in the 2 ha site. Tree size The median value was computed for each wood-pasture based on the

diameter at breast height (DBH) in cm of trees higher than 3 m. These median values across all wood-pastures were then averaged.

Scrub cover The percent cover of scrub in the wood-pasture. Scrub was defined in our study as vegetation dominated by woody perennials (shrubs and young trees), usually exceeding the height of the grass layer, and being between 0.2 m and ca 3 m in height. Characteristic shrub species were: Hawthorn (Crataegus monogyna), Blackthorn (Prunus spinosa), Blackberry (Rubus sp.) and the Dog Rose (Rosa canina). The most common young tree was the Hornbeam (Carpinus betulus).

Woody vegetation cover The percent coverage of woody vegetation (trees and shrubs) in the whole wood-pasture. Source: data derived from a supervised classification of the monochromatic channels of SPOT 5 data (©CNES 2007, Distribution Spot Image SA) using a support vector machine algorithm (Knorn et al. 2009).

(b) Composition variables Prevalence of Oak, Hornbeam, Beech and Pear

Defined as the proportion of Oak, Hornbeam, Beech and Pear in relation to the complete number of trees measured in a given site.

(c) Site related variables Area The size of the wood-pasture (ha). Source: satellite imagery and GIS. Elevation In meters (m). Source: recorded in situ with a Global Positioning System. Ruggedness The ruggedness of the terrain for the whole wood-pasture and was

calculated as the standard deviation of elevation in a 25 m x 25 m grid (Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Elevation Model Version 2 (GDEM V2))

(d) Landscape related context Forest cover Percentage of forest cover within a 300 m buffer from the edge of wood-

pasture based on CORINE Land Cover classes. Source: European Environmental Agency (2011): (http://www.eea.europa.eu/publications/COR0-landcover).

Distance to nearest village The Euclidean distance from the centre of the wood-pasture to the edge of the closest village calculated in GIS (in m).

(e) Management related variables Scrub cleaning (APIA) The presence of scrub clearance. Cut scrub collected in piles was

considered evidence for APIA activities at the site. Livestock Cattle, sheep, buffalo, horses or a mixture of these. The percent of

wood-pastures grazed by each of these livestock was calculated. Burning Presence of burning in the wood-pasture.


141

Analysis

Raw data were summarized using descriptive statistics. Due to the different scaling of the data the

Coefficient of Variation (CV) was used as dimensionless measure of variability, allowing a

meaningful comparison among different variables. Illustration of tree communities was based on

genera composition from forests versus wood-pastures using detrended correspondence analysis

(DCA), using the number of stems belonging to different genera. The DBH of burned and dead

trees was compared against that of healthy trees using t-tests. Prior to this, DBH data were log-

transformed to meet assumptions about the distribution of the data. Density of dead trees per

hectare was compared between forests and wood-pastures using a t-test.

To model indicators of condition and composition we used an information-theoretic

model selection approach based on the Akaike information criterion (AIC) to identify models best

supported by the data (Burnham & Anderson 2002). We separately considered six different

response variables: the proportion of dead trees and the proportion of scrub cover were variables

from the ‘condition’ group (see above and Table 6.1). The prevalence of Oak, Hornbeam, Beech

and Pear were ‘compositional’ variables (see above and Table 6.1). We selected the condition

variables as responses because (i) they directly influence the quality of the (wood-) pasture (scrub

cover, proportion of dead trees); and (ii) they are highly dynamic variables – both increased after

the 1989 Romanian revolution, but since the entry of Romania into the European Union (2007

hereafter EU), financial incentives (i.e. APIA payment, see above) have been used to clean pastures

of scrubs. We selected the compositional variables because these tree species were abundant and

some of them (e.g. Oak, Pear) historically have a strong cultural importance for local communities

(Dorner 1910). Each response variable (see above) was modelled separately as a function of six

explanatory variables from the following categories (detailed in Table 6.1): ‘Site related’ (S) variables

(altitude and ruggedness), ‘Management related’ (M) variables (evidence of sheep grazing or not,

evidence of ‘APIA’ related scrub removal) and ‘Landscape related’ (L) variables (forest cover and

distance to nearest village).

We constructed seven candidate models arising from all combinations of the groups of

explanatory variables listed above (M, S, L, M + S, M + L, S + L, M + S + L). All continuous

variables were standardized to an average of zero and a standard deviation of one in order to make

the effects comparable. For each model, the AIC value was calculated using correction for small

samples sizes (AICc, Burnham and Anderson, 2002). The models were ranked according to their

AICc, where the best model has the smallest AICc value. Delta AICc (D AICc) was calculated to

express the difference between each model and the best model. Akaike weights (w) were used to

estimate the relative evidence for each model, which could be interpreted as the probability that the

model i was the best model for the observed data, given the candidate set of models.

Chapter VI

142

Results

Tree community composition in wood-pastures and forests

We measured 6739 trees, including 4870 in forests and 1869 in wood-pastures. Twelve tree genera

were found in forests and 14 in wood-pastures (Table 6.2). Tree communities differed between

wood-pastures and forests (Fig. 6.2). Forest sites were dominated by Hornbeam (Carpinus betulus)

and Beech (Fagus sylvatica), while wood-pastures were dominated by Oak (Quercus sp.) (Fig. 6.2).

Table 6.2 Tree genera identified in the forest and wood-pasture sites. The fourth column shows which

species are known to occur in our region for each genus (based on Coldea 1992). A plus denotes presence,

minus denotes absence.

Genus Forest Wood-pasture Species from the region

Acer + + Acer campestre, A. platanoides, A. pseudoplatanus, A. tataricum, A. negundo3

Betula + + Betula pendula

Carpinus + + Carpinus betulus

Fagus + + Fagus sylvatica

Fraxinus + + Fraxinus excelsior

Juglans - + Juglans regia

Larix + + Larix decidua1

Malus - + Malus domestica2

M. sylvestris2

Pinus + + Pinus nigra3

P. sylvestris3

Populus + +

Populus alba P. tremula

Prunus + + Prunus avium, P. spinosa P. cerasifera 2 P. domestica2

Pyrus - + Pyrus communis2

P. pyraster2

Quercus + + Quercus petraea Q. pubescens Q. robur Q. robur × Q. petraea Quercus rubra3

Robinia + + Robinia pseudoacacia1

Salix - + Salix alba S. caprea, S. cinerea, S. fragilis, S. purpurea, S. triandra

Tilia + + Tilia cordata


143

Figure 6.2 Tree community composition in forests versus wood-pastures. A) Illustration of tree

communities based on genera composition from forests versus wood-pastures using detrended

correspondence analysis. Eigenvalues are given in brackets for each axis. The length of first axis is 3.1 and

that of the second axis is 1.7. B) Percentage of the most common tree genera found in forests and wood-

pastures.

Smaller (and presumably younger) trees were better represented in forests than in wood-

pastures, while larger (presumably older) trees were better represented in wood-pastures (Fig. 6.3).

Wood-pastures contained more ancient trees and more trees of conservation value, while only two

individuals of such trees were found in forests (Table 6.3). Dead trees were observed in every forest

site but 64% of wood-pastures contained no dead trees. The average number of dead trees per

hectare for the remaining wood-pastures was 2 (min–max: 1–6) while forest sites contained on

average 7 (min–max: 1–16) dead trees per hectare, this difference being significant (t-test, P <

0.05).

Chapter VI

144

Table 6.3 The number of ancient trees and trees of conservation value found in the wood-pastures (n = 42)

and forests (n = 15). The most common tree genera are presented. Numbers in brackets are sites (i.e. wood-

pastures) where trees from that category were observed.

Truly ancienta (DBH ≥ 2 m)

Conservation valuea (DBH ≥ 1.5 m)

Potentially interestinga

(DBH ≥ 1 m) Ancientb

Wood-pastures

Oak (n = 1113) 3 (2) 73 (11) 175 (32) 251 (31)

Hornbeam (n = 255) 1 (1) 0 9 (4) 37 (16)

Beech (n = 100) 0 0 23 (8) 0

Forests

Oak (n = 883) 0 0 1 1

Hornbeam (n = 1994) 0 0 0 1

Beech (n = 1782) 0 0 0 0 a Read (2000) b Farm Environment Plan Guide (2006): DBH ≥ 75 cm for Hornbeam, ≥ 100 cm for Oak, ≥ 150 cm for

Beech

Figure 6.3 The percent representation of the tree size categories in forests and wood-pastures. Only Oak,

Hornbeam and Beech are shown because these were the most common trees present in both forests and

wood-pastures.


145

The characteristics and current management of wood-pastures

The 42 wood-pastures covered a total area of 42.21 km2. Descriptive statistics describing wood-

pastures are presented in Table 6.4. Sixty percent of wood-pastures were grazed only by sheep,

21% only by cattle and 14% were grazed by a mixture of livestock (i.e. cattle, sheep, buffalo) (Table

6.4). Evidence for scrub removal (APIA) was found in 88% of wood-pastures (Table 6.4).

Management by burning in 2012 was observed in 50% of the wood-pastures (Table 6.4).

Comparison of the size of burned versus unburned trees was possible only for Oak and Hornbeam

due to insufficient sample sizes for other species. Burned Oaks (mean DBH = 123.93, SD = 39.99,

n = 76) and dead Oaks (mean DBH = 123.28, SD = 48.73, n = 13) were significantly larger than

healthy Oaks (mean DBH = 83.56, SD = 32.59, n = 1014) (t-tests, P < 0.001 and P < 0.05

respectively; 10 trees injured by humans and struck by lightning were not included). Burned

Hornbeams (mean DBH = 70.95, SD = 28.51, n = 10) were larger than unburned Hornbeams

(mean DBH = 53.03, SD = 26.03, n = 243 two dead trees were not included) (t-test, P < 0.05).

Across all wood-pastures sampled (though not necessarily within our survey sites), we observed

that more than 40 ancient Oaks (sensu Farm Environment Plan Guide 2006) collapsed following

uncontrolled pasture fires in 2012. In 2012 we recorded tree (Oak and Hornbeam) cutting activities

in five wood-pastures, two of them being ancient (sensu Farm Environment Plan Guide 2006).

Table 6.4 Descriptive statistics for variables used to characterize the wood-pastures. CV = coefficient of

variation.

Variable Mean SD CV

Condition variables Tree density 7.60 4.80 0.63

Number of scattered trees 260.64 230.44 0.88

Proportion of dead trees 0.06 0.10 1.72

Tree size 74.25 23.11 0.31

Scrub cover 0.06 0.12 1.90

Woody vegetation cover 0.19 0.14 0.73

Site related variables Area 100.50 90.97 0.90

Elevation 543.65 68.21 0.12

Ruggedness 62.75 14.54 0.29

Landscape context Forest cover 61 0.23 0.37

Distance to nearest village 1225.17 804.17 0.65

Management related variables Scrub cleaning (APIA) Observed in 88% of wood-pastures

Livestock Cattle: 21%, Buffalo: 7% (always mixed with

other livestock), Sheep: 60%, Mixed: 14.28. No grazing: 5%

Burning 50%

Chapter VI

146

Models of wood-pasture condition and composition

For condition-related response variables, the models best supported by the data contained either

management variables (M; scrub cover) or variables describing the landscape context (L;

proportion of dead trees) (Table 6.5). For the prevalence of particular genera, the best models

always included all three groups of variables (M + S + L) (Table 6.5). GLM analysis showed that

the proportion of dead trees was positively related to surrounding forest cover (Table 6.6).

Moreover the prevalence of Oak, Hornbeam and Beech was positively and the prevalence of Pear

was negatively associated with forest cover (Table 6.6). Oak prevalence was negatively and Beech

prevalence was positively related to ruggedness (Table 6.6). Distance to the nearest village was

related negatively to the prevalence of Pear and positively to the prevalence of Beech (Table 6.6).

Oak prevalence was negatively and Pear prevalence was positively associated with elevation (Table

6.6).

Table 6.5 Model selection results for response variables describing different aspects of wood-pasture

condition and composition. The best ranked models (∆i ≤2) are shown. Log(L) = maximised log-likelihood,

K = number of estimable parameters, ∆I = difference in AICc compared with the model with the lowest

AICc, wi = Akaike weights; S = site variables, M = management variables, L = landscape variables. Variables

are defined in the Methods section.

Table 6.6 The relationship between the response variables (first column) and the explanatory variables,

separately tested using GLMs.

Livestock type

APIA clearing

Altitude Ruggedness Forest cover

Village distance

Proportion of dead trees

NS NS NS NS 0.47 (0.21)* NS

Scrub cover NS NS NS NS NS NS Oak prevalence

NS -0.50 (0.19) -0.19 (0.06)* -0.45 (0.06)** 0.23 (0.06)***

NS

Hornbeam prevalence

-0.92 (0.16)***

0.66 (0.23)** NS 0.24 (0.06)*** 0.67 (0.09)***

NS

Beech prevalence

-1.26 (0.26)***

1.61 (0.53)** NS 0.56 (0.08)*** 0.66 (0.16)***

0.30 (0.13)*

Pear prevalence

0.56 (0.16)***

0.70 (0.28)* 0.39 (0.07)***

NS -0.61 (0.08)***

-0.17 (0.08)*

NS = non-significant.

*** = P<0.001.

** = P<0.01.

* = P≤0.05.

Response variables Model Log (L) K AICc Δi wi

Proportion of dead trees

L -71.14 4 151.37 0.00 0.60

Scrub cover M 38.53 4 -67.98 0.00 0.54

Oak prevalence M+S+L

-254.81

7

526.92

0.00

0.66

S+L -258.34 5 528.36 1.43 0.32 Beech prevalence M+S+L -114.72 7 246.47 0.00 0.99 Hornbeam prevalence M+S+L -173.62 7 364.54 0.00 0.98 Pear prevalence M+S+L -169.81 7 356.92 0.00 1.00


147

Discussion

In this paper we showed that forests and wood-pastures differed with respect to their tree

community structure, typical tree sizes and the prevalence of dead trees. We also showed that sheep

grazing dominated wood-pastures while the other livestock (cattle, horse and buffalo) were scarcely

used. Burning as a management tool was widely applied, and large trees appeared to be particularly

affected by this. Scrub clearance induced by the EU level financial incentives was applied in most

wood-pastures. Finally, the prevalence of different species and of dead trees in wood-pastures was

related to management, site and landscape related variables.

Wood-pastures versus forests

While wood-pastures were dominated by Oak and fruit trees (mostly Pear), forests had a more

balanced proportion of Beech, Oak and Hornbeam. Differences between the tree communities of

forests and wood-pastures can be explained the ecology of the trees (Vera 2000), natural

prerequisites and the traditional preferences of local people for Oak and fruit trees. The potential

primary vegetation in the study region is represented by mixed Oak and Hornbeam, and mixed

Beech and Hornbeam forests. Mixed Oak and Hornbeam forests (Quercus petraea, Q. robur, C.

betulus) were found on shaded and semi shaded hills while mixed Beech and Hornbeam forests (F.

sylvatica, C. betulus), have a more zonal distribution on valley slopes (Coldea 1992). Historical

information suggests that many wood-pastures from Southern Transylvania originate from forest

grazing and selective tree removal from forests (Teşculă & A. 2007; Hartel & Moga 2010).

Transylvanian Saxons traditionally valued Oak not only for timber production but also (and

especially) for the acorns, which were eaten by domestic pigs and sometimes sheep (Dorner 1910;

Oroszi 2004). The importance of grazing for wood-pasture formation and their maintenance is well

known for other European wood-pastures (Mountford & Peterken 2003).

Our results highlight that the largest trees in Southern Transylvania occurred in wood-

pastures: the majority of the surveyed wood-pastures contained ancient Oaks while forest sites

contained virtually no such trees. Within the same bioclimatic conditions large trees of a given

species are typically older than smaller ones (Gibbons & Lindenmayer 2003; Holzwarth et al. 2013),

and hence the relative proportion of young trees appeared to be higher in forests than in wood-

pastures. Age estimations for Oaks from the ‘Breite’ ancient wood-pasture (situated close to the

centre of the study region) suggest that a tree with a DBH of ≥100 cm may be at least 200 years

old, and the largest Oaks may be up to 700–800 years old (Hartel & Moga 2010; Patrut 2011). The

main reason for these size (and age) differences could be the long term management of forests and

wood-pastures. In our region trees were maintained on pastures mostly to provide shade for

livestock and for their fruit (TH, unpublished results of 110 semi structured interviews in Southern

Transylvania). Timber extraction occurred also in wood-pastures, but it was done mainly by

pollarding (i.e. cutting branches while maintaining the trunk) (Rackham 1980; Hartel & Moga

2010). This allowed trees to grow and eventually to become large (old). By contrast, forests were

Chapter VI

148

traditionally managed for timber production (see Oroszi 2004 for and overview of forest

managament by Transylvanian Saxons), and the economic value (in a monetary sense) of trees was

and still is important in determining management actions. Old trees were and are removed from

forests because their economic value is decreasing as the amount of dead elements and hollows

increase with age. Assuring sustainability of the forests by regeneration (naturally ‘from seed’ or by

replanting the cleared parcels) was important both traditionally (Oroszi 2004) and continues to be

common practice nowadays (Codul Silvic (Forest Code of Romania), Law 46/2008). However, no

mechanisms to re-plant trees (or otherwise support their regeneration) occur in the vast majority of

wood-pastures.

Management of wood-pastures

Sixty percent of wood-pastures surveyed in 2012 were grazed only by sheep and 14% by a mixture

of livestock (Table 6.4). This is in sharp contrast with traditional grazing systems: Saxons preferred

cattle, horses, buffalo and pigs, and typically each of these had its own pastures with specific

management practices around a given village (Dorner 1910; and also Heinlein et al. 2005 for

selected areas of Bavaria, Germany). Pig grazing stopped in the late 1940s and 1950s, and the

number of buffalo dropped after the Romanian revolution in 1989, partly because of mass

emigration of Saxons and partly for economic reasons. Cattle grazing also declined sharply in many

wood-pastures after 1989 (although less so than for buffalo), and the number of sheep is now

higher than ever before in the Saxon region of Transylvania (TH, unpublished results of 110 semi

structured interviews in Southern Transylvania). Changes to the traditional grazing systems also

have been reported from many European wood-pastures throughout Europe (Plieninger & Schaar

2008; Bergmeier et al. 2010; Costa et al. 2011; Garbarino et al. 2011; Chételat et al. 2013).

Half of the wood-pastures were burned in 2012, and large trees (Oak and Hornbeam)

appeared to be most likely to be permanently damaged from this. Fire has been used as a method

for pasture clearing in the region since the 16th century (Dorner 1910). However, uncontrolled

pasture burning appears to have increased in recent years, even in protected areas, despite being

illegal (TH, personal observation).

Wood-pasture condition

Our model selection approach showed that the prevalence of Oak, Beech, Hornbeam and Pear

trees in wood-pastures was best explained by models containing management, site and landscape

related variables. Oak dominated in wood-pastures with low terrain ruggedness while Beech

dominated in areas with high ruggedness. This result can be explained partly by the ecology of

these species (see above) and partly by human influence. For example, it is possible that Oak was

retained in flatter terrain where accessibility for livestock was high and which contained wet areas

often preferred by these animals (especially cattle, domestic pigs and buffalo). Fruit trees (especially

Pear) dominated in wood-pastures close to villages (suggesting that accessibility for people was


149

important in creating them) with little forest cover in their surroundings. By contrast, forest cover

was an important positive predictor for the abundance of all three forest tree species. A likely

explanation is that many, if not most, thinning existing forests created wood-pastures in our region.

The pear prevalence was positively associated with the elevation; this is most likely a result of the

slight increase of the average elevation toward the Eastern part of the region, where this tree is

more abundant. Our result regarding the significant relationship found between the management

related variables and prevalence of different tree genera (Table 6.5) is not straightforward: the

relationship is likely caused by the fact that local conditions shaped human activity (this being

recorded in 2012) and not the other way round. For example, it is possible that wood-pastures

where beech and hornbeam dominated were less attractive for grazing, possibly because of steeper

slopes or higher woody vegetation density. The grazing system in our region is undergoing rapid

changes, and therefore this relationship is likely to change in the future.

The percentage of dead trees was significantly related to forest cover around the wood-

pasture. Wood-pastures surrounded by forests may be less accessible for people than the wood-

pastures from open landscapes. Traditional rural communities from this region carefully cleared the

pastures of dead wood and scrub to maintain pasture quality (Dorner 1910; TH, unpublished

results of 110 semi structured interviews in Southern Transylvania). Based on this, it is reasonable

to assume that wood-pastures in traditional societies contained dead wood only accidentally and if

the scrub was present, it was deliberately maintained (e.g. as occasional firewood or as source of

fruits). It is possible that the increase of the dead wood on pastures is the result of the

abandonment of pastures, which was very pronounced after the 1989 Romanian revolution. As the

continuation of use of pastures and hay meadows is promoted by EU agro-environment incentives,

it is likely that abundance of dead wood will decline again in the future in most of wood-pastures

from this region.

The biodiversity of the ‘Breite’ ancient wood-pasture

In the context of wood-pasture conservation, it is important to note that some wood-pastures in

our study area have been shown to support a very rich diversity of plants and animals. A wide

range of studies have been conducted on one of the wood-pastures also surveyed by us for this

paper, namely the ‘Breite’ wood-pasture, situated near the centre of our study area, near the town

Sighisoara. The Breite measures 133 ha and is completely surrounded by deciduous forest. It is

dominated by Oaks (mostly Q. robur), many of which are over 200 years old (Patrut 2011). Overall,

476 species of vascular plants (Öllerer, 2012), 121 species of macromycetes (from which over 50

species were found on ancient Oaks - Bucşa 2007; Bucşa & Tăuşan 2010), 281 species of

Lepidoptera, 40 species of xylophagous beetles (i.e. insects to which wood represents the primary

diet), eight species of amphibians, four species of reptiles, 27 species of nesting birds and 38 species

of mammals (including Gray Wolf, Canis lupus, and Brown Brear, Ursus arctos) have been identified

in this wood-pasture (synthesized in Hartel & Moga 2010; Hartel et al. 2011). The overall number

Chapter VI

150

of species considered rare or protected at national (e.g. Red List) and international (e.g. IUCN,

Habitats and Birds Directives) level exceeds 50.

Conclusions and conservation implications

We showed that there were differences between the tree communities in wood-pastures and

forests. Ancient trees were found only in wood-pastures, and most of the surveyed wood-pastures

contained ancient trees. Historical and current management, traditional preferences of local people,

and natural environmental gradients are likely explanations for these differences. Fire appears to be

regularly used in pasture management, but our data suggest that uncontrolled fires can negatively

affect (or even kill) trees, especially large ones. Further, our data, in combination with historical

records, suggest that major changes are underway regarding patterns of livestock grazing in

Southern Transylvania, implying that the management of wood-pastures is shifting from traditional

practices. Demographic and economic factors are the likely drivers of these recent shifts. Data

from one of the wood-pastures in the centre of our study area suggest that the presence of

scattered, old trees, in combination with dead trees, scrub and extensively managed grassland,

results in a high biodiversity, with species rich communities of woodland and grassland related

organisms. To maintain the ecological value of wood-pastures, at least some of the dead trees and

scrub need to be maintained - although this runs counter to both traditional practices and current

policy incentives. Wood-pastures are currently managed as pastures (or occasionally as hay

meadows) and are formally recognized as such at the national level (Romanian Law No. 214/2011).

While tree cutting from wood-pastures without a formal institutional agreement is illegal in

Romania (Law 214/2011), there is no legal framework that specifically targets the maintenance and

regeneration of wood-pastures and the conservation of old (including ancient and veteran) trees.

With very few exceptions, wood-pastures are not recognized in the nature conservation policies of

the EU and are not protected as distinct land cover types with special management

history, ecological and cultural value. Therefore their maintenance as such is not promoted at

policy level. Our study shows that ancient wood-pastures are common in our region, and we

suggest that they also may be common in other CEE countries. We urge for more wood-pasture

inventories and research in other parts of CEE, to develop the knowledge base that is needed for

their formal recognition and legal protection.


151

Acknowledgements

The early stages (i.e. 2010 and before) of wood-pasture research in Southern Transylvania were

founded by the Mihai Eminescu Trust, EEA Grants through the project ‘Conservation of the

biodiversity of the Breite Ancient Oak tree nature reserve, Sighisoara’ and ‘The implication of local

communities in the conservation of wood-pasture habitats from the Saxon villages of

Transylvania’. The present research and the research stay of TH in Germany were supported by

Alexander von Humboldt Foundation. JF was funded by Alexander von Humboldt Foundation via

a Sofja Kovalevskaja Award. We thank Árpád Szapanyos for his help with the tree measurements.

Appendix A. Supplementary material

Figure 1. KML-file of the surveyed wood-pasture sites shown in Google Earth Map

Figure 2. KML-file of the surveyed forest sites shown in Google Earth Map

152

153

Chapter VII

154

155

Chapter VII

The conservation value of traditional rural landscapes:

The case of woodpeckers in Transylvania, Romania

I. Dorresteijn, T. Hartel, J. Hanspach, H. von Wehrden, J. Fischer

PLoS ONE (2014), 8:e65236

‘When a dead tree falls, the woodpeckers share in its death.’ – Malayan proverb quote

Chapter VII

156

Woodpeckers in wood pastures

157

Abstract

Land-use change is a major threat to global biodiversity. Forest species face the dual threats of

deforestation and intensification of forest management. In regions where forests are under threat,

rural landscapes that retain structural components of mature forests potentially provide valuable

additional habitat for some forest species. Here, we illustrate the habitat value of traditional wood

pastures for a woodpecker assemblage of six species in southern Transylvania, Romania. Wood

pastures are created by long-term stable silvo-pastoral management practices, and are composed of

open grassland with scattered large, old trees. Because of their demanding habitat requirements,

woodpeckers share habitat with many other bird species, and have been considered as possible

indicator species for bird species diversity. We first compared woodpecker assemblages between

forests and wood pastures. Second, we grouped features of wood pastures into three spatial

contexts and addressed how these features related to the occurrence of three woodpecker species

that are formally protected. Woodpecker species composition, but not the number of species,

differed between forests and wood pastures, with the green woodpecker occurring more commonly

in wood pastures, and the lesser spotted woodpecker more commonly in forests. Within wood

pastures, the intermediate context (especially surrounding forest cover) best explained the presence

of the grey-headed and middle spotted woodpecker. By contrast, variables describing local

vegetation structure and characteristics of the surrounding landscape did not affect woodpecker

occurrence in wood pastures. In contrast to many other parts of Europe, in which several species

of woodpeckers have declined, the traditional rural landscape of Transylvania continues to provide

habitat for several woodpecker species, both in forests and wood pastures. Given the apparent

habitat value of wood pastures for woodpeckers we recommend wood pastures be explicitly

considered in relevant policies of the European Union, namely the Habitats Directive and the EU

Common Agricultural Policy.

Chapter VII

158

Introduction

Human-induced landscape change poses a major threat to global biodiversity (Fahrig 2003; Foley et

al. 2005). Forest species face the dual threat of deforestation and intensification of forest

management. Woodpeckers are especially sensitive to these changes because they require large

home ranges and depend on large trees for nesting and dead wood for foraging (Angelstam &

Mikusinski 1994). Consequently, changes in forest structure and cover have caused woodpecker

declines worldwide (Conner & Rudolph 1991; Roberge et al. 2008; Lammertink et al. 2009).

Because of their demanding habitat requirements, woodpeckers share habitat requirements with

many other bird species. Therefore, woodpeckers have been considered as potential indicator

species for bird species diversity (Roberge & Angelstam 2006).

In Europe, six out of ten species of woodpeckers are protected under the EU Birds

Directive Annex I. While several woodpecker species have declined in Western Europe, Eastern

European woodpecker assemblages have remained diverse and stable due to the persistence of

large forest tracts (Mikusinski & Angelstam 1998). However, Eastern European forest landscapes

are increasingly coming under pressure from more intensive and widespread logging operations

(Knorn et al. 2012). Against this background, it is important to understand to what extent the

landscape context surrounding forest patches could provide complementary habitat for

woodpeckers in Eastern Europe. Many agricultural landscapes in Eastern Europe are characterized

by low-intensity subsistence farming that still contain semi-natural vegetation cover, including

transitional elements between forest patches such as scattered trees (Tryjanowski et al. 2011). Such

relatively complex landscapes have the potential to support high biodiversity. However, the links

between biodiversity and land-use are still relatively poorly understood in Eastern Europe

(Tryjanowski et al. 2011). Mikusinski and Angelstam (1998) proposed that rural landscapes

retaining remnant structures of natural forests could provide valuable additional habitat for species

otherwise confined to forests, but little work has specifically tested this claim to date.

One of the most intact traditional rural landscapes in lowland Europe occurs in southern

Transylvania (Romania). In this area, traditional wood pastures are of particular interest in terms of

their potential habitat value for woodpeckers. Many wood pastures in southern Transylvania are

several centuries old; they are composed of open grassland with scattered trees and result from

ancient silvo-pastoral management practices. Wood pastures bring together three ecologically

important components. First, scattered trees have a disproportionate ecological value, that is their

effect on ecosystem functioning is disproportionately large relative to the small area occupied by an

individual tree, both locally (Fischer et al. 2010) and at a landscape scale (Manning et al. 2006;

Manning et al. 2009). Second, many wood pastures contain large and old trees (Robles et al. 2007;

Hartel & Moga 2010), and therefore incorporate structural attributes that support biodiversity

elements typical of old-growth forests (Bauhus et al. 2009). Third, wood pastures can provide

habitat for both open-country and forest species. Despite their high potential to support

biodiversity, the ecological importance of wood pastures, especially in Eastern Europe, remains


159

poorly understood (but see e.g. Mountford & Peterken 2003; Taboada et al. 2006; Bergmeier et al.

2010). However, existing evidence from Western Europe suggests that wood pastures can provide

habitat for a range of bird species including woodpeckers and secondary cavity-nesting birds

(Robles et al. 2007; Robles et al. 2011).

Here, we document the habitat value of wood pastures for an assemblage of six

woodpecker species in Transylvania, Romania. Our specific objectives were to (1) compare

woodpecker communities between wood pastures and forests; and (2) assess which features of a

wood pasture are particularly important for different woodpecker species. Our findings highlight

the conservation value of wood pastures in traditional rural landscapes in Eastern Europe.

Methods

Ethics statement

All necessary permits were obtained for the described study. Permission to survey woodpeckers

within the EU Natura 2000 network was granted by Progresul Silvic, the organisation officially

entrusted with the custody of the protected area by the Romanian government. The survey

procedure, including the use of playback calls for protected species, was cleared by the ethics

committee of Leuphana University Lueneburg.

Woodpeckers were surveyed in 28 wood pastures and 12 forests in Southern Transylvania,

Romania (Fig. 7.1a). We sampled a larger number of wood pastures than forests because forests are

relatively homogenous, whereas wood pastures differ substantially in structural elements and

adjacent forest cover. Furthermore, we were especially interested in which features of a wood

pasture affected woodpecker presence and thus chose more sites to comprehensively cover existing

gradients within wood pastures. Wood pastures were chosen on the basis of availability and access,

and to cover a gradient in surrounding forest cover (min. = 3.3%; max. = 96.8%; mean ± SE =

59.7 ± 4.8%). Within each wood pasture we choose one survey site, located approximately in the

centre of the wood pasture, for woodpecker point counts. Forests were chosen on the basis of

accessibility, and each forest site was located randomly at a distance of at least 600 m from the

forest edge. Wood pastures and forests differed in tree species composition: forests were

dominated by oak (Quercus robur and Q. petrea), hornbeam (Carpinus betulus) and beech (Fagus

sylvatica), whereas wood pastures were dominated by oak and had more fruit trees (mainly pear,

Pyrus pyraster) (Table 7.1). Trees were larger (and typically older) and occurred at a lower density in

wood pastures than in forests (Table 7.1).

Chapter VII

160

Figure 7.1 Study area and design. A) The location of the study area in southern Transylvania, Romania, and

location of the 12 forest and 28 wood pasture sites. B) Example of a survey site, showing the three different

landscape extent groups considered in the analyses. The small dashed circle (a 300 m radius around the

survey point) represents the local context; solid lines represent the intermediate context (a 300 m buffer from

the border of the wood pasture); and the large dashed circle represents the broader landscape context (a 2000

m radius around the survey point).

Table 7.1 Habitat characteristics (mean ± SE) of the two surveyed habitat types: forests (n = 12) and wood

pastures (n = 28). The number of trees was calculated as the number of trees in 2 ha; median dbh (cm) was

calculated as the mean of the medians measured within 2 ha and between 80 and 300 m; the proportion of a

tree species was calculated as the mean of the proportion of a species in 2 ha and between 80 and 300m.

We surveyed the six most common woodpecker species, namely great spotted woodpecker

(Dendrocopos major), middle spotted woodpecker (D. medius), lesser spotted woodpecker (D. minor),

green woodpecker (Picus viridis), grey-headed woodpecker (P. canus), and black woodpecker

(Dryocopus martius). The middle-spotted woodpecker, grey-headed woodpecker and black

Habitat variables; mean ± SE Forest Wood pasture

Number of trees 1345 ± 255.89 16.54 ± 1.77

Median DBH 25.97 ± 2.69 76.52 ± 4.86

Proportion of oak 0.20 ± 0.03 0.63 ± 0.04

Proportion of hornbeam 0.44 ± 0.05 0.12 ± 0.03

Proportion of beech 0.33 ± 0.07 0.05 ± 0.02

Proportion of fruit trees 0.003 ± 0.001 0.15 ± 0.03


161

woodpecker are protected under the EU Birds Directive Annex I. We did not include the Syrian

woodpecker (Dendrocopos syriacus) and white-backed woodpecker (Dendrocopos leucotos) into our study

because they are very rare in our study area, and we did not encounter these species during our

surveys. We used unlimited point counts enhanced by the use of playbacks of woodpecker

drummings and/or calls with an audible range of approximately 150 m. Following 5 minutes of

initial listening, playbacks were used in a sequence from smallest to largest species (Kumar & Singh

2010). For each species, a sequence of drummings and/or calls was played three times for 15

seconds, with 15 seconds of silence between playbacks and 1.5 minutes of silence between species.

All sites were surveyed three times during appropriate weather conditions between 10 March and 6

April 2012.

To characterize the structure of sites, we recorded the species and diameter at breast height

(dbh) of: (1) within 80 m of the survey point (= 2 ha), all trees in wood pastures (1 to 34 trees) or

50 trees in forests (selected randomly in a spiral from the survey point to the edge of 2 ha); and (2)

in four strip transects of 10 m width between 80 and 300 m from the survey point in the four

cardinal directions (N, S, E, W). In case that no or too few trees were found on a transect we

measured between one and four trees close to the transect. For forest sites, we estimated tree

density in 2 ha by doubling the count of all trees in two opposite quarters of a 2 ha circle; in wood

pastures we counted all trees within 2 ha around the survey point.

All analyses were performed on presence/absence data, pooled across the three repeats,

and implemented in the ‘R’ environment (R Development Core Team 2013). First, we assessed

differences in woodpecker species composition and the number of species between forest sites and

wood pastures. We conducted nonmetric multi-dimensional scaling based on Sørensen dissimilarity

to assess the differences in woodpecker composition between forests and wood pastures using the

R package ‘Vegan’. We used analysis of similarity (ANOSIM; Sørensen dissimilarities) with 1000

permutations to test for differences in woodpecker composition between forests and wood

pastures. In a last step, we modelled the number of woodpecker species as a function of site type

using a generalized linear model with Poisson error structure. Although using playbacks and three

repeat surveys presumably decreased the incidence of false absences, they cannot be ruled out. This

is especially the case for the species that were relatively uncommon (number of times a species was

observed at least twice compared to total number of sites present: great spotted woodpecker:

21/33; middle spotted woodpecker: 7/15; lesser spotted woodpecker: 1/9; grey-headed

woodpecker: 9/22; green woodpecker: 23/30; black woodpecker: 4/10). Importantly, however, we

are confident that detectability did not differ between forests and wood pastures. Potential

differences in visibility between forests and wood pastures were of minor importance because: (1)

most woodpeckers were identified using calls; and (2) trees did not yet have leaves and visibility in

forests was therefore good.

Second, we assessed which features of a wood pasture were important for different species

of woodpeckers. The response of the three species protected under the EU Birds Directive to

Chapter VII

162

environmental variables was analysed using generalized linear models with a binomial error

structure. We did not include the forest sites in these analyses because their habitat characteristics

were very different from wood pastures (Table 7.1), and we were specifically interested in which

features of wood pastures affected woodpecker occurrence. We did not analyse the three

unprotected species because they occurred too rarely or too frequently to be modelled (site

occurrence of great spotted woodpecker: 23/28; green woodpecker: 26/28; lesser spotted

woodpecker: 5/28)

We hypothesized that different site-specific and landscape variables might influence

woodpecker presence, and we therefore grouped our explanatory variables as follows: local context

(L) represented by a circle with a 300 m radius around a site; intermediate context (I) represented as

an irregular buffer of 300 m around a particular wood pasture; and broader landscape context (B)

represented by a circle with a 2000 m radius around a site (Fig. 7.1b). Scales were selected both to

match the scale of the wood pastures and to be ecologically meaningful to woodpeckers.

Local variables included the proportion of oak, median dbh, and percent woody vegetation

cover. Woodpeckers generally prefer older trees for nesting, with several species showing a

particular preference for oak (Kosiński & Kempa 2007; Barrientos 2010). Proportion of oak was

estimated by calculating the mean of the proportion of oak in 2 ha and between 80 and 300 m, and

median dbh was estimated by calculating the mean of the median diameters measured within 2 ha

and between 80 and 300 m. Percent woody vegetation was derived from a supervised classifications

of the panchromatic channels of SPOT 5 data (©CNES 2007, Distribution Spot Image SA) using a

support vector machine algorithm (Knorn et al. 2009).

Intermediate context variables were related to specific structures of the wood pasture and

included the area of the wood pasture and percent woody vegetation cover within a 300 m buffer

from the edge of the wood pasture (as a measure of how much of the perimeter of a given wood

pasture was adjacent to forest). Adjacency to forest cover was considered important because it may

positively or negatively affect the use of wood pastures by woodpeckers.

Landscape variables related to compositional heterogeneity and terrain ruggedness within a

radius of 2000 m around the survey point. This radius corresponds to the approximate average

valley width of the study area. Birds occurring in farmland mosaics have been observed to respond

strongly to landscape heterogeneity (Benton et al. 2003). Compositional heterogeneity was

calculated as the standard deviation of the monochromatic channel of SPOT 5 data (©CNES 2007,

Distribution Spot Image SA). Terrain ruggedness was calculated as the standard deviation of the

elevation. It indicated the geomorphology of the surrounding landscape, and also functioned as a

proxy for forest cover because highly rugged landscapes tended to be densely forested.

Prior to modelling we log transformed percent woody vegetation within 300 m of the

survey point, as well as area of the wood pasture; we standardized all variables by subtracting their

mean and dividing by their standard deviation; and we confirmed that variables were not correlated.

We then used an information theoretic approach for model selection to identify models that best


163

explained woodpecker presence (Burnham & Anderson 2002). We constructed eight alternative

candidate models arising from all possible combinations of the three groups of variables (I, I+L,

I+B, I+L+B, L, L+B, B) and the null model. We used the R package ‘AICmodavg’ to rank the

candidate models, based on AICc values to account for small-sample bias (Burnham & Anderson

2002). Models considered best had an AICc difference (∆AICc) of less than two from the model

with the lowest AICc.

Results

We found differences in species composition between wood pastures and forests in the ordination

and the analysis of similarity (NMDS: two axes, stress = 15.1, see Fig. 7.2; ANOSIM: R = 0.141, P

= 0.009). The R-statistic from the analysis of similarity was only slightly larger than zero indicating

that compositional dissimilarity between groups was only slightly larger than within groups. Only

two species showed a clear habitat preference: the green woodpecker for wood pastures and the

lesser spotted woodpecker for forests (Fig. 7.2). The number of species ranged from two to six in

wood pastures, and from zero to four in forests. The mean number of species did not differ

significantly between wood pastures (mean: 3.14 ± 0.25) and forests (mean: 2.58 ±0.38) (GLM, z =

0.91, P = 0.34).

We detected the middle-spotted woodpecker in 11, the grey-headed woodpecker in 16, and

the black woodpecker in 7 of our 28 wood pasture sites. The best ranked models for individual

species included the intermediate context group and the null model for the middle spotted

woodpecker, and the intermediate context and broader landscape context groups for the grey-

headed woodpecker (Table 7.2). Within the intermediate context, the amount of woody vegetation

within 300 m from the edge of the wood pasture (i.e. a proxy of the amount of perimeter that was

forested) had the largest effect on both species, with both responding positively to a more forested

perimeter (Table 7.3). For the black woodpecker only the null model was selected as the best-

ranked model (Table 7.2).

Chapter VII

164

Figure 7.2 Woodpecker species composition in forests and wood pastures. Non-metric multidimensional

scaling of woodpecker composition in both forests and wood pastures, based on a Sørensen dissimilarity

matrix (two axes; stress = 15.1).


165

Table 7.2 Full model summary of all the candidate models for the three woodpecker species protected under

the EU Bird Directive Annex I.

Best ranked models (Δi < 2) are marked with *. Model: L = local context; I = intermediate context; B = broader landscape context; Null = null model

Model summary: Log(L) = the maximised log-likelihood, K = number of estimated parameters; AICc:

Akaike’s Information Criterion corrected for small sample bias; ΔAICc: difference in AICc compared with

the model with the lowest AICc; Wi: Akaike weights.

Species Model Log(L) K AICc Δ AICc Wi

Middle spotted

woodpecker I* -15.69 3 38.39 0.00 0.479

Null* -18.76 1 39.67 1.29 0.252

B -16.86 3 40.73 2.34 0.149

B+I -15.05 5 42.83 4.44 0.052

L -16.77 4 43.28 4.89 0.041

L+B -14.28 6 44.58 6.19 0.021

L+I -15.73 6 47.47 9.08 0.005

L+I+B -13.94 8 51.46 13.07 0.001

Grey-headed

woodpecker I* -15.06 3 37.12 0.00 0.606

I+B* -13.12 5 38.97 1.85 0.241

Null -19.12 1 40.40 3.27 0.118

B -18.94 3 44.87 7.75 0.013

L+I -14.65 6 45.05 7.92 0.012

L -18.05 4 45.54 8.42 0.009

L+I+B -12.41 8 48.39 11.26 0.002

L+B -17.95 6 51.65 14.53 0.000

Black woodpecker Null* -15.75 1 33.64 0.00 0.701

L -13.91 4 37.56 3.91 0.099

I -15.33 3 37.66 4.02 0.094

B -15.60 3 38.21 4.56 0.072

L+B -12.65 6 41.31 7.67 0.015

I+B -14.52 5 41.76 8.12 0.012

L+I -13.56 6 43.12 9.48 0.006

L+I+B -11.62 8 46.76 13.12 0.001

Chapter VII

166

Table 7.3 Model coefficients (and standard errors) of the environmental variables included in the best

ranked models in binomial GLMs for the middle spotted and grey-headed woodpecker.

Model: I = intermediate context; B = broader landscape context; Null = null model

Discussion

Our findings highlighted that traditional wood pastures, as well as forests, provide useful habitat

for woodpeckers in southern Transylvania. The value of wood pastures was particularly evident for

the green woodpecker, which was more likely to occupy wood pastures than forests. This finding is

consistent with earlier work by Rolstad et al. (2000) who suggested that home ranges of the green

woodpecker were often confined to meadows and pastures. The diet of the green woodpeckers

consists of ants, and it actively selects foraging sites with a high ant biomass (Rolstad et al. 2000;

Alder & Marsden 2010). High ant abundance occurs in grazed semi-natural grasslands (Dauber et

al. 2006), whereas forests typically support lower ant abundances than open areas (Rolstad et al.

2000). Furthermore, the green woodpecker avoids foraging in areas with tall and dense vegetation

(Alder & Marsden 2010). Thus, the structure and management of wood pastures support optimal

foraging habitat for the green woodpecker (and possibly for other species specialized on ants).

The lesser-spotted woodpecker was more strongly associated with forests than wood

pastures. Although the home ranges of the lesser spotted woodpecker sometimes include open

areas, the species typically avoids open areas for foraging (Wiktander et al. 2001). The lesser spotted

woodpecker feeds on insects such as aphids, beetle larvae and ants, which are often found in dead

wood (Angelstam & Mikusinski 1994). This suggests that wood pastures could provide potential

foraging habitat for the lesser spotted woodpecker, and foraging requirements therefore cannot

explain why the species appeared to avoid wood pastures. A possible alternative explanation is that

the lesser-spotted woodpecker may avoid wood pastures to reduce predation risk. Other authors

have suggested that the lesser spotted woodpecker appears to select locations with lower predation

risk (e.g. forests compared to more open wood pastures) over locations with higher energetic

profitability (Olsson 1998 cited in Wiktander et al. 2001).

However, wood pastures may be used by other woodpecker species typically associated

with forest environments, as demonstrated in Spanish dehesas for the middle spotted woodpecker

(Robles et al. 2007). Indeed, we found no difference in the number of species between wood

Model terms

Species Model (Intercept) Size wood

pasture

Woody

vegetation in

perimeter

Hetero-

geneity

Ruggedness

Middle spotted

woodpecker I -0.58 ± 0.45 0.40 ± 0.48 1.03 ± 0.52

Null -0.44 ± 0.39

Grey-headed

woodpecker I 0.34 ± 0.45 -0.25 ± 0.43 1.31 ± 0.55

I + B 0.48 ± 0.51 -0.47 ± 0.51 2.41 ± 0.98 0.001 ± 0.51 -1.27 ± 0.72


167

pastures and forests, and all three woodpecker species protected by the EU Bird Directive Annex I

(and typically considered to be forest-associated) were present in both types of sites.

Surprisingly, environmental variables at different spatial scales had little effect on the

presence of protected woodpecker species. The only environmental variable positively related to

the presence of two species was surrounding forest cover (especially evident for the grey-headed

woodpecker; Table 7.3). The grey-headed woodpecker may preferentially occupy forest stands

containing beech (Shurulinkov et al. 2012), which in our study area more frequently occurred in

forests rather than wood pastures. It is possible, therefore, that the grey-headed woodpecker selects

nest sites in beech trees within forest patches (Kosiński & Kempa 2007) but uses nearby wood

pastures for foraging. We observed the grey-headed woodpecker twice or more in only 9 out of the

22 sites in which it was ultimately detected, which may support the notion that it uses wood

pastures for foraging rather than breeding. The grey-headed woodpecker forages on ants, although

it is less specialized compared to the green woodpecker (Angelstam & Mikusinski 1994) – it is

primarily a ground feeder when the ground is free of snow or forages on bark-dwelling insects on

dead trees during winter (Rolstad & Rolstad 1995). While it appears that wood pastures should

provide good foraging habitat for the grey-headed woodpecker, data on breeding locations would

be required to further scrutinize this explanation of occurrence patterns.

Similarly to the grey-headed woodpecker, the black woodpecker also may preferentially

select beech trees for nesting (Kosiński & Kempa 2007), and its diet also includes ants (Angelstam

& Mikusinski 1994). However, the black woodpecker is likely to use the landscape at a different

spatial scale compared to the other two species. It moves over larger areas (Tjernberg et al. 1993)

and the intermediate spatial scale chosen in our study may have been too small for this species to

show an effect of adjacent forest cover.

Local vegetation structure, including the availability of large trees, is known to influence

woodpecker presence elsewhere (e.g. Pasinelli 2000). The lack of association of woodpecker species

with local variables in our study may reflect that wood pastures contain enough old trees to provide

critical habitat elements such as dead wood and food resources (Angelstam & Mikusinski 1994) -

the amount of old-growth elements thus apparently did not limit woodpecker distribution.

Habitat fragmentation by loss of forests has been hypothesized to be one of the major

causes of forest bird declines (Robinson et al. 1995). Because woodpeckers have large home ranges,

they may be highly sensitive to forest fragmentation (Pettersson 1985; Wiktander et al. 1992).

Given the widespread occurrence of woodpeckers throughout our study area, our results suggest

that woodpeckers are likely to perceive the landscape as largely unfragmented: (1) we found no

difference in the number of species between forests and wood pastures; and (2) environmental

variables such as ruggedness (a proxy for landscape level forest cover) had little effect on the three

threatened woodpecker species. Broad-leaved forest cover in our study area is approximately 42%,

and existing evidence suggests that fragmentation effects becomes severe only well below this

threshold (Tjernberg et al. 1993; Andrén 1994; Wiktander et al. 2001). For example, the black

Chapter VII

168

woodpecker was largely insensitive to fragmentation in a highly fragmented landscape with only

26% of forest cover (Tjernberg et al. 1993). Moreover, scattered trees are available throughout the

agricultural mosaic, very likely providing effective functional connectivity between forest patches

(Manning et al. 2006). Because we lack information on reproductive performance of woodpeckers

in wood pastures compared to forests, we cannot make inferences about the quality of wood

pastures as breeding habitat for woodpeckers. Nevertheless, our results suggest that wood pastures

provide important feeding habitat for woodpeckers and probably provide connectivity between

different forest patches.


Despite their high cultural and natural values, wood pastures are declining rapidly throughout

Europe (Bergmeier et al. 2010). Considering their important values for a range of species (Taboada

et al. 2006; Paltto et al. 2011; Robles et al. 2011; Horák & Rébl 2013), the conservation of wood

pastures should be addressed in relevant policies and directives. For woodpecker conservation

specifically, we recommend that conservation policies focus not only on maintaining mature forests

but also recognize the complementary value of wood pastures that have retained old-growth

structures. There are a few national conservation policies for wood pastures (e.g. Goldberg et al.

2007), but to date there is no pan-European conservation policy (Bergmeier et al. 2010).

To improve the conservation status of European wood pastures we suggest they should be

considered explicitly in two major EU policies: (1) the EU Habitats Directive; and (2) the EU

Common Agricultural Policy, specifically with respect to agri-environment payments. Currently,

wood pastures are inconsistently considered in the EU Habitats Directive, and the Directive does

not include Romanian wood pastures (Bergmeier et al. 2010). Although the ecological effect of

some agri-environment payments is still debated (Kleijn et al. 2006), when appropriately targeted,

they can provide a useful tool for farmland biodiversity conservation (Donald & Evans 2006). In

many member states of the EU, agri-environment payments exist for management of extensive

pastures, which typically provide subsidies for clearing woody vegetation to maintain extensive

grassland environments. However, such payments may inadvertently pose a threat to wood

pastures, because scattered tress can fall victim to the clearing process (legally or accidentally, pers.

obs.; (Blom 2012)). More carefully specified agri-environment schemes could stimulate the

conservation of trees in extensively managed pastures, thereby recognizing their keystone role for a

wide range of organisms (Manning et al. 2006).

Acknowledgements

We thank Arpad Szapanyos and Anne-Catherine Klein for their tremendous effort in the field. We

thank Patrick Griffiths, Jan Knorn and Tobias Kümmerle for their help with creating the woody

vegetation layer. We thank four anonymous reviewers for constructive suggestions on a previous

draft of this manuscript.

169

Chapter VIII

170

171

Chapter VIII

Brown bear activity in traditional wood-pastures in Southern

Transylvania, Romania

M. Roellig, I. Dorresteijn, H. von Wehrden, T. Hartel, J. Fischer

Ursus (2014), 25:43-52

‘We're going on a bear hunt!

We're gonna catch a big one!

I'm not afraid!

Are you?

Not me!’

– Children’s song,

Michael Rosen

Chapter VIII

172

Brown bear activity in wood pastures

173

Abstract

During the past century, habitat fragmentation and increased human impacts have reduced

populations of large carnivores throughout the world. Although bears have been extirpated from

human-dominated landscape in most parts of Europe, they still occur in multiuse cultural

landscapes in Southern Transylvania, Romania. Wood-pastures - grazed permanent grasslands with

scattered or clumped trees and shrubs - are important elements of these cultural landscapes and

provide habitat for a wide range of species. However, wood-pastures are under threat from land-

use change, including intensified agricultural use and land abandonment. In 2012 we assessed the

level of activity of brown bears (Ursus arctos) and environmental factors influencing bear activity in

54 wood-pastures in Southern Transylvania. As an index of bear activity, we measured the

proportion of anthills that were destroyed by bears. The variables were combined in 3 groups

(anthropogenic effects, local variables, and landscape context) to test which group most strongly

influenced bear activity. Bear activity was found in 47 (87%) wood-pastures. Bear activity was best

explained by variables describing the landscape context, with proximity to the Carpathian

Mountains, terrain ruggedness, and amount of surrounding woody vegetation positively related to

bear activity. Local variables (distance to forest edge and anthill density) had no effect, and

surprisingly, variables related to anthropogenic features (distance to major roads, distance to

villages) were positively related to bear activity (albeit weakly). Most of the wood-pastures in

Southern Transylvania were used by bears for foraging on ant larvae. For the conservation of

brown bears in Southern Transylvania, it is important to maintain large areas of forest but also

consider cultural landscape elements such as wood-pastures. To conserve European wood-

pastures, we suggest they be explicitly considered in national nature conservation policies and in

major European Union (EU) policies such as the EU Habitats Directive.

Chapter VIII

174

Introduction

Habitat fragmentation and increased human impact have reduced populations of large carnivores

throughout the world (Breitenmoser 1998; Woodroffe 2000). The brown bear (Ursus arctos) is

Europe’s largest carnivore, but has been displaced from human-dominated landscapes in most

parts of Europe (Swenson et al. 1995; Zedrosser et al. 2011; Kopatz et al. 2012). Most remaining

populations are confined to mountainous regions and heavily depend on major conservation

efforts (Swenson et al. 2000; Zedrosser et al. 2001). Thus, the need to protect the brown bear in

Europe remains a priority (Zedrosser et al. 2011; Krofel et al. 2012), even though the species has

been listed in Annex II and Annex IV of the European Union (EU) Habitats Directive

(92/43/EEC) since 1992. Although bear populations are now increasing in some areas (Zedrosser

et al. 2011), agricultural intensification and habitat fragmentation, resource extraction, road

development, recreation development, and urban expansion still threaten Europe’s remaining bear

populations (Swenson et al. 1995; Breitenmoser 1998; Zedrosser et al. 2001).

Romania has the largest national brown bear population in Europe (excluding Russia;

Chestin 1999; Spassov & Spiridonov 1999), estimated by the International Union for Conservation

of Nature, to be stable at approximately 6,000 individuals (Huber 2007). In Transylvania, bears

occur not only in large forest patches, but also in multi-use cultural landscapes. However, potential

landscape changes such as intensified agriculture, deforestation, and planned new highways could

negatively affect Transylvania’s bear population (Swenson et al. 2000; Ministry of Agriculture

Forestry and Rural Development and the Ministry of Environment and Water Management

[MAPDR and MMGA] 2005).

Changes in land-use also pose a major threat in multi-use cultural landscapes such as the

traditional grazing systems of wood-pastures (Plieninger 2007; Bergmeier et al. 2010). Wood-

pastures are grazed permanent grasslands with scattered trees and shrubs or with groups of trees

and shrubs, and important elements of cultural landscapes in Europe (Bergmeier et al. 2010).

Wood-pastures are used by a wide range of species, including many threatened species (Tucker &

Evans 1997; Wegener 1998; Olea & San Miguel-Ayanz 2006; Bergmeier et al. 2010; Dorresteijn et

al. 2013). Plants, invertebrates, and birds are the most commonly studied species in wood-pastures

but, apart from some publications mentioning use of the Spanish dehesas by Iberian lynx (Lynx

pardinus; Olea & San Miguel-Ayanz 2006; Bergmeier et al. 2010), relatively little is known about the

use of wood-pastures by large carnivores. To our knowledge, this is the first ecological study

focusing on the use of wood-pastures by large carnivores.

Wood-pastures in Southern Transylvania (Fig. 8.1) are generally open, with a density of

approximately 8 trees/ha (Hartel et al. 2013). Some wood-pastures are surrounded by forest, while

others are situated within open pastures or, more rarely, arable fields. Wood-pastures in Southern

Transylvania are mainly grazed by sheep, cattle, and sometimes goats and buffalos in moving herds.

The main tree species are oak (Quercus spp.) and beech (Fagus sylvatica), but also fruit trees such as

wild pear (Pyrus silvestris) and hornbeam (Carpinus betulus). Characteristic shrub species are hawthorn


175

(Crataegus monogyna), blackthorn (Prunus spinosa), blackberry (Rubus spp.) and dog rose (Rosa canina;

Hartel et al. 2013). Most of these trees provide potential food sources (mast and fruits) for brown

bears.

Anthills are also typical elements of woodpastures, and the larvae within these provide an

important source of protein for brown bears in spring (Swenson et al. 1999). The most common

antspecies in grasslands (such as the wood-pastures in Southern Transylvania) include yellow

meadow ants (Lasius flavus), black garden ants (L. niger), L. paralienus, and pavement ants

(Tetramorium cf. caespitum; B. Markó and K. Erös, Babes-Bolyai University, Cluj-Napoca, Romania,

unpublished data). Three of these (L. flavus, L. niger, and T. cf. caespitum) are probably consumed by

bears, based on the abundance of nests in the field, the abundance of individuals in the nests, and

the amount of larvae and pupae (B. Markó , personal communication).

Wood-pastures in Southern Transylvania thus contain several useful elements for brown

bears: (1) their heterogeneous character provides a wide variety of food sources including ant larvae

in spring, fruits and berries in the summer, and hard mast during autumn; (2) groups of trees or

shrubs provide immediate refuge for bears (Swenson et al. 2000); and (3) they are used extensively

by grazing livestock in moving herds (Hartel et al. 2013), meaning that pastures are not fenced and

are accessible for bears.

To date, it is largely unknown which environmental variables explain bear activity in

cultural landscapes such as wood-pastures, and at which scales these variables are important. To fill

this knowledge gap, we investigated bear activity in wood-pastures in Southern Transylvania. We

aimed to ascertain (1) whether, and to what extent, wood-pastures were used by brown bears; and

(2) which environmental variables, and at which scale, could explain bear activity in wood-pastures.

Figure 8.1 Typical wood-pasture in southern Transylvania, Romania.

Chapter VIII

176

Methods

Study area and site selection

We conducted the study in Southern Transylvania, Romania. Southern Transylvania is a plateau

surrounded by the Carpathian Mountains. We investigated bear activity in 54 wood-pastures

around the town of Sighisoara (Fig. 8.2). The study area was chosen because of the widespread

occurrence of wood-pastures and available information about them (Hartel & Moga 2010).

We defined wood-pastures as pastures with scattered trees and shrubs, or with groups of

trees and shrubs (Bergmeier et al. 2010). To locate wood-pastures, we used published literature

(Hartel & Moga 2010) as well as CORINE pasture distribution data (CORINE landcover;

European Environment Agency [EEA] 2006) and satellite images of the area (Google Earth). We

selected wood-pastures with a broad gradient in percentage surrounding forest cover. Within each

wood-pasture, we established 2–6 transects, depending on the size of the wood-pasture (<30 ha =

2 transects; 30–80 ha = 3 transects; 80–130 ha = 4 transects; 130–180 ha = 5 transects; >180 ha =

6 transects). At each site, we placed one transect 7 m from, and parallel to, the forest edge and one

in the centre of the pasture. Additional transects were placed randomly via Geographic Information

System and were typically separated by several hundred meters. Each transect was 400 m long and

6 m wide.

To quantify bear activity, we documented the proportion of anthills disturbed by foraging

activity (i.e., we counted all destroyed and intact anthills in each transect). This method was

developed and successfully used by the Mammal Conservation Workgroup of the ‘‘Milvus Group’’

Bird and Nature Protection Association for monitoring bear activity in Transylvania

(http://milvus.ro/Mammal_Conservation). In each transect, we noted the coordinates in longitude

and latitude using a Global Positioning System at 100-m intervals, and recorded the number of

intact anthills (diam > 25 cm, ht > 15 cm) and destroyed anthills in the previous 100-m segment.

All surveys were conducted from 1 May to 24 June 2012.

Figure 8.2 Study area in southern Transylvania, showing Romania, the Carpathian Mountains, forest areas,

villages, major roads, and locations of the 54 wood-pasture sites.


177

Environmental variables

Bears are known to react to both human activity (e.g. Clevenger et al. 1992; Nellemann et al. 2007;

May et al. 2008) and environmental conditions (Slobodyan 1974; Swenson et al. 2000), and we

assumed based on their large range size (Preatoni et al. 2005) that they are sensitive to multiple

spatial scales. Therefore, to investigate which combination of variables best explained bear activity

in wood-pastures, we grouped our explanatory variables as follows: (1) anthropogenic variables

(AV), (2) local variables (LV), and (3) variables describing the (broader) landscape context (LC).

Each of the variables was calculated based on the midpoints of each 100-m segment.

We used anthropogenic variables to describe disturbance, measured as the shortest

distance to village limits and major roads, based on each midpoint. For local variables, we estimated

the proportion of woody vegetation cover within a 300-m-radius circle and measured the distance

to forest edge (using CORINE forest data from 2006 [EEA 2006]), as a proxy for distance to

shelter, from each segment’s middle point. We also accounted for possible effects of the total

number of anthills/100 m by including them as a local variable, because bears might be more

attracted to pastures with more anthills. Finally, landscape-context variables included woody

vegetation cover and terrain ruggedness within a 3-km radius around each segment’s midpoint, as

well as distance to the Carpathian Mountains, measured from each segment’s midpoint to the edge

of the Carpathian Mountains. Both woody vegetation cover and ruggedness may be important with

respect to the availability of sheltering habitat and a wider variety of food for bears. Woody

vegetation and distance to the Carpathian Mountains may be important for connectivity to a larger

(source) population and to dens at the higher altitudes. We derived woody vegetation from a

supervised classification of the monochromatic channels of SPOT 5 data (ECNES 2007;

Distribution Spot Image SA, Toulouse, France) using a support vector machine algorithm (Knorn

et al. 2009). We calculated terrain ruggedness as the standard deviation of the elevation (Advanced

Spaceborne Thermal Emission and Reflection Radiometer [ASTER] Global Digital Elevation

Model Version 2 [GDEM V2]; Land Processes Distribution Active Archive Center, Sioux Falls,

South Dakota, USA).

Modelling bear activity

As an index of bear activity, we quantified the proportion of destroyed anthills relative to the total

number of anthills in each 100-m segment of each transect. The resulting data structure therefore

was nested, with 4 segments/transect, and several transects per site (i.e., pasture). Thus, we

modelled activity at the transect level to simplify the structure of random effects to 2 nested levels

(transect/site). We used a binomial error distribution for the response data. All statistical analyses

were implemented in the R environment (R Development Core Team 2013), and generalized linear

mixed models were fitted using the package ‘lme4’ (Bates et al. 2014).

To account for the nested data structure, we initially included site as a random intercept in

the model, while in a second model we accounted for potential changes in activity over time by

Chapter VIII

178

including Julian date as a random slope. Using both date and site as random factors at the same

time was not meaningful because they were redundant, meaning each site was sampled on a specific

date. Because site explained a greater amount of variance and arrived at more parsimonious models

based on initial Akaike Information Criterion (AIC) comparisons, we used the model that included

site only as a random intercept.

Distance to forest and anthill density were strongly skewed; therefore, we log-transformed

these variables before modelling. Furthermore, we standardized all predictors by subtracting their

mean and dividing by their standard deviation to make the estimates comparable. Because of a

strong correlation between distance to forest edge and woody vegetation within a 300-m-radius

circle (Spearman rank correlation; r = -0.68), we removed woody vegetation from the group of

local variables.

We used an information theoretic approach for model selection to identify models that

best explained bear activity (Burnham & Anderson 2002). We constructed 7 alternative candidate

models, which considered all possible combinations of the 3 groups of variables (AV, LV, LC). We

used the R package AICmodavg (Mazerolle 2012) to rank the candidate models, based on AICc

values to account for small-sample bias (Burnham & Anderson 2002). The AICc values of the best

models were compared with the AICc values of the null model (which only included the random

effects). Marginal and conditional R2 values were calculated following the procedure outlined by

Nakagawa and Schielzeth (2013), using the package ‘arm’ (Gelman et al. 2013).

Results

The size of wood-pastures ranged from approximately 11 ha to 475 ha, but most were 30 - 130 ha

(<30 ha, n = 8; 30–80 ha, n = 19; 80–130 ha, n = 15; 130–180 ha, n = 6; >180 ha, n = 6). We

surveyed 199 transects, with a mean of 3.6 transects/wood-pasture. Anthills were found in all

pastures; only 1 of the transects and 21 of the 100-m segments contained no anthills.

We detected destroyed anthills in 87% of wood-pastures and in 42% of transects. We

found 29,236 total anthills, including 1,771 destroyed anthills (6%). Within a given 100-m segment,

we recorded up to 300 anthills, and up to 127 destroyed anthills (mean total no./100 m ≈ 37; mean

destroyed/100 m ≈ 2).

Mean distance to nearest village was 2 km (range = approx. 150 m to approx. 5 km). The

distance to the major roads averaged 35 km (120 m to 100 km). The average distance to the

Carpathian Mountains was 30 km (11 - 64 km) and average distance to forest edge was 0.2 km (7 m

to 1.8 km). The woody vegetation cover within 3,000 m averaged 44% (7% - 84%). Terrain

ruggedness averaged 57 m (32 - 105 m). All models including explanatory variables performed

better than the null model (Table 8.1). The best supported model (wi = 0.66) included

anthropogenic variables (AV) and landscape context variables (LC). Despite the low difference in

AIC values (<2) between the best supported and the second-ranked model, we selected the most

parsimonious one (i.e., containing AV and LC), because the contribution of the third group of


179

variables (local variables; LV) was unlikely to contribute meaningful explanatory power to the

model (Burnham & Anderson 2002; Arnold 2010), given the low estimates of the local variables

(Table 8.2). The marginal (i.e., only explaining the fixed effects) R2 value for the top model was

low, but the conditional R2 value (explaining the random and fixed effects) was 70% (Table 8.1).

Table 8.1 Summary of candidate models influencing brown bear activity in wood-pastures, southern

Transylvania, Romania, including the maximised log-likelihood (Log(L)), number of estimable parameters

(K), AICc, difference in AICc compared with the model with the lowest AICc (∆i), and Akaike weights (w i).

Also shown are the marginal and conditional R² values for all models. (AV = Anthropogenic variables, LV =

Local variables, LC = Landscape context).

Landscape context was the most important variable group related to bear activity. Within

this group, the distance to the Carpathians Mountains had the strongest effect. Bear activity

increased with decreasing distance to the mountains. Also terrain ruggedness and percentage of

woody vegetation cover in the surrounding area were positively related to bear activity in wood-

pastures (Table 8.2). Bear activity was slightly correlated with the anthropogenic variables: we

found higher bear activity in wood-pastures that were closer to villages, but almost no relationship

between bear activity and proximity to roads.

Model Summary

Model Log(L) K AICc Δi Wi Marginal R² Conditional R²

AV+LC -1778.81 7 3571.77 0.00 0.66 0.125 0.695

AV+LV+LC -1777.46 9 3573.16 1.39 0.33 0.117 0.686

LC -1785.53 5 3581.15 9.38 0.01 0.179 0.754

LV+LC -1784.27 9 3582.68 10.90 0.00 0.167 0.7434

AV+LV -1801.77 6 3615.64 43.87 0.00 0.017 0.451

AV -1804.16 4 3616.36 44.60 0.00 0.014 0.446

LV -1811.37 4 3630.78 59.00 0.00 0.003 0.429

Null Model -1813.90 2 3631.81 60.04 0.00

Chapter VIII

180

Table 8.2 Best-supported model describing which variables explained bear activity in wood-pastures,

southern Transylvania, Romania. The table shows the model coefficients (and standard errors) that were

included in the best ranked models. Also shown is the second ranked model to highlight the low estimates

for both variables of the group ‘local variables’ (AV = Anthropogenic variables, LV = Local variables, LC =

Landscape context).

Discussion

Our findings show that most of our studied wood-pastures in Southern Transylvania were used by

brown bears, confirming the relevance of non-forest habitats (Slobodyan 1974; Swenson et al.

2000). In the following, we reflect on the survey method used, the variables related to bear activity,

and potential conservation implications arising from our work.

Survey method

Ants are a well-known food resource for bears (Swenson et al. 1999), and several previous studies

have used destroyed anthills as a sign for the presence of bear activity (Munro et al. 2006; Ciarniello

et al. 2007). Using the proportion of destroyed anthills to total anthills turned out to be a

pragmatic, but apparently useful, index of bear activity in our study, despite the unequal

distribution of anthills across wood-pastures. Ideally, we would have compared our activity

estimation with some other measure of bear activity known to be reliable. No such validation was

possible during the study period; however, a similar study in 2012 in the same study area also used

destroyed anthills as an index for bear activity and generated ecologically consistent results (I.

Dorresteijn et al., Leuphana University, Lueneburg, Germany, unpublished data). Studies on the

foraging behavior of bears have shown that bears often target ants during spring when numbers of

pupae increase (Noyce et al. 1997; Swenson et al. 1999). Because of the harsh winter of 2011 -

2012, ant larvae were available later in the season than expected, and bear foraging activity

increased rapidly a few weeks after we started our surveys. Additionally, it is plausible that activity

may increase over time because we considered anthills during a period of 2 months. Both of these

issues could result in a positive effect of time on observable foraging activity; however, such effects

were corrected for by including site (and thus implicitly survey date) into our model as a random

intercept. We did not take into account potential selection by bears for different ant species, but we

acknowledge that this may occur and warrants addition research.

Model terms

Model (Intercept) Distance to villages (km)

Distance to roads (km)

% Woody vegetation 3km

Terrain Rugged-ness

Distance to mountains (km)

Distance to forest edge (km)

Anthill density (no./0.06ha)

AV+LC 3.67±0.33 -0.26±0.07 -0.05±0.06 0.29±0.10 0.74±0.11 -0.90± 0.30

AV+LC+LV

3.72±0.34 -0.26±0.07 -0.05±0.6 0.30±0.10 0.72±0.11 -0.82±0.30 -0.02±0.03 -0.14±0.09


181

Variables influencing bear activity in wood-pastures

Bear activity in wood pastures was primarily driven by environmental variables at the landscape

level and to a lesser extent by anthropogenic variables. The importance of the landscape variables is

consistent with previous studies, which also identified mountainous regions (Posillico et al. 2004),

as well as rugged and forested areas, as being preferred by bears for shelter, denning sites, and food

(Clevenger et al. 1992; Preatoni et al. 2005; Nellemann et al. 2007; May et al. 2008). The relative

lack of importance of local variables was surprising because we hypothesized that distance to the

forest edge could be an important element of escape over (Swenson et al. 2000) and expected

higher bear activity closer to forests. However, the relatively small size and complex shape of many

of the studied wood-pastures may have ensured that the forest edge was generally within close

proximity. Alternatively, the structurally rich wood-pastures could have provided adequate shelter.

Anthill density also did not appear to influence bear activity; fewer anthills than were available in

our study area are potentially still sufficient in number to attract bears.

Bear occurrence has been shown to be negatively affected by human presence (Clevenger

et al. 1992; Preatoni et al. 2005; Nellemann et al. 2007; May et al. 2008). However, bear activity in

our study was slightly positively associated with the anthropogenic variables we measured. This

unexpected finding may be explained by most villages in Southern Transylvania being small and

characterized by low human density (approx. 34 inhabitants/km2; Arion et al. 2011). Elsewhere in

Europe, bears inhabit areas with human densities up to 80 people/km2 (Linnell et al. 2001). In

Scandinavia, female brown bears avoided human activity, but used human-dominated landscapes

for foraging when human density was low, especially at night (Martin et al. 2010). Such temporal

niche partitioning between humans and bears may also occur in Southern Transylvania. Major

roads were generally several kilometers from the wood-pastures we studied, which may explain

their lack of influence on bear activity.

Conservation and management implications

Although bears typically require large, relatively undisturbed areas, we showed that they also use

wood-pastures within the cultural landscapes of Southern Transylvania. This is likely because they

represent a source of protein (ants), which is critically important in the otherwise mainly vegetarian

diet of the brown bear in temperate latitudes (Bojarska & Selva 2012). Wood-pastures likely

provide fruits and hard mast as food sources for brown bears in summer and autumn, but our

investigations were confined to the spring season when ants and their larvae were most abundant

and sought after by bears.

Wood-pastures in Southern Transylvania face several threats. Land abandonment leads to

forest succession, which will eventually displace species linked to the semi-open character of wood-

pastures, including fruit trees and invertebrates such as the grassland ant species (both food sources

for bears). On the other hand, intensified agricultural use, including clearing of tree cover or

fertilization, will destroy the semi-open and semi-natural character of the wood-pastures, which is

Chapter VIII

182

also a threat to animal and plant species (Manning et al. 2006). Additional threats to wood-pastures

are illegal burning in spring, which often leads to the destruction of trees, and legal and illegal

cutting of trees. These factors, in combination with occasional overgrazing and clearing the

pastures of shrubs as part of the requirements for EU agricultural subsidies, leads to a lack of

regeneration (Hartel et al. 2013). Within Europe, some types of wood-pastures (such as the dehesas

in Spain and the Fennoscandia wooded pastures in Scandinavia) are protected under the EU

Habitats Directive, but there is no consistent conservation approach across Europe (Bergmeier et

al. 2010). A loss of Transylvania’s wood-pastures could mean the loss of some food sources for

bears, but also the loss of a buffer between forest patches and villages, which could increase

conflicts between people and bears. The protection and management of wood-pastures in Southern

Transylvania thus could make a positive contribution to bear conservation.

Finally, a critical aspect of bear conservation in Southern Transylvania is the maintenance

of a network of large and well-connected forest patches. Southern Transylvania has approximately

30% forest cover, which offers several potential corridors to nearby mountains and source

populations. We did not observe a negative effect of roads on bear use of wood-pastures.

However, other studies have shown the negative impact of highways, and newly planned major

highways could negatively impact bear activity by causing habitat fragmentation (Kaczensky et al.

2003; Karamanlidis et al. 2011). Further research is needed on the importance of connectivity of

forest patches to the mountains and how it may be affected by improvements and/or alterations to

the local road network.

Zedrosser et al. (2011) suggested that bear conservation in Europe should not be restricted

to only wild and remote areas, because few such areas remain (Linnell et al. 2005), and because

other landscape elements can provide useful complementary conservation opportunities. Our

findings suggest this is the case for our study area, where bears also use wood-pastures within

multi-use cultural landscapes.

Acknowledgements

We are deeply grateful to the team of the Milvus Mammal Conservation group, especially A.

Kecskés, H. Latková, S. Szilárd, and Z. Mezey for establishing and sharing with us the survey

method described in this paper. For creating the woody-vegetation layer we thank P. Griffiths, J.

Knorn, and T. Kümmerle. For his opinion on ant consumption by bears we thank Balint Markó.

We thank also A. Szapanyos and M. Solomon for their efforts in the field, and L. Sutcliffe for

providing language assistance and for proof reading the manuscript. Lastly we thank the

anonymous reviewers for constructive suggestions on a previous draft of this manuscript. The

project was funded by the Alexander von Humboldt Foundation through a postdoctoral fellowship

to T.H. and a Sofja Kovalevskaja Award to J.F.

183

Section C: Human-

bear coexistence

This Section includes Chapters IV (see Section A) and IX and takes a social-ecological systems

approach to understand how links between the social and ecological parts of the system affect

human-bear coexistence. Chapter IV aims to obtain an overview on human-bear conflicts in the

study region. Chapter IV also examines whether the level of perceived conflict relates to bear

activity and whether it influences attitudes of shepherds towards bears. Chapter IX aims to gain a

more holistic understanding on the social drivers shaping the willingness of people to coexist with

bears.

184

185

Chapter IX

186

187

Chapter IX

Social factors mediating human-carnivore coexistence:

Understanding coexistence pathways in Central Romania

Ine Dorresteijn, Andra Ioana Milcu, Julia Leventon, Jan Hanspach, Joern Fischer

In revision in AMBIO

‘Nature is not a place to visit, it is home.’

– Gary Snyder

‘Well, for us the bear is like our neighbor.

Like the neighbor at home. There's no

difference between my neighbor and the bear

that comes every second evening. We're

somehow used to him.’ – Shepherd from

Vidacut, Southern Transylvania

Chapter IX

188

Drivers of human-carnivore coexistence

189

Abstract

Successful carnivore conservation depends on the degree to which humans and carnivores can

coexist. Facilitating human-carnivore coexistence does not only depend on the biophysical

environment but also on understanding the social factors that shape human-carnivore coexistence.

Focusing on Central Romania, we aimed to understand social drivers of coexistence by applying a

combined approach of questionnaires (n = 252) and inductive semi-structured interviews (n = 70)

to examine human-bear (Ursus arctos) coexistence. The majority of respondents had a positive

perception of coexistence. The questionnaires revealed general patterns on the most important

social drivers, and highlighted four distinct groups of respondents. Perceptions of coexistence were

negatively influenced by negative attitudes towards bears, past negative interactions with bears, and

perceived risks of damage. Qualitative content analysis of interviews identified three coexistence

pathways. These pathways showed different (but not mutually exclusive) ways in which ongoing

interactions between the ecological system and the social system shape the willingness of people to

coexist with bears. While the connection between humans and the landscape had important

mediating effects on perceived coexistence, the generally positive perception of coexistence could

be undermined by perceived flaws in official bear management. We suggest that the concept of

coexistence pathways, coupled with a comprehensive approach to their analysis, can help landscape

managers identify drivers that facilitate or hinder coexistence, and thus target interventions

accordingly. This conceptual framework, in turn, may help facilitate human-carnivore coexistence

worldwide.

Chapter IX

190

Introduction

Where humans and large carnivores share the same landscape, there are inevitable conflicts:

humans experience attacks and predation on livestock, with resultant economic impacts (Thirgood

et al. 2005; Holmern et al. 2007); and carnivore populations decline as a result of persecution and

growing human pressure on carnivore habitat (Woodroffe 2000; Ripple et al. 2014). Therefore,

successful carnivore conservation depends not only on the biophysical environment but also on

understanding the social factors that shape human-carnivore coexistence (Treves & Karanth 2003).

Importantly, human tolerance towards carnivores is not only shaped by the experience of damage

(Hazzah et al. 2009; Dickman et al. 2014; Kansky et al. 2014). Rather, it is constructed through a

variety of factors related to economic, aesthetic, ecological, cultural, religious, and intrinsic values

ascribed to carnivores (Zinn et al. 2000; Dickman 2010). For example, traditional and cultural

differences between pastoralist and agriculturalists lead to differences in their tolerance towards

lions in South Africa (Gusset et al. 2008; Lagendijk & Gusset 2008). Moreover, in many places the

political environment has added an additional factor influencing human-carnivore interactions, for

example through the implementation of top-down conservation management, financial incentives,

or tight legislation (Redpath et al. 2012), which may clash with the priorities of rural populations

(Skogen et al. 2008; Majić et al. 2011).

To design effective tools that facilitate coexistence, studies need to account for the

complexity of social factors that shape it (Dickman 2010). Despite an increasing recognition of the

need to integrate social science into understanding the extent of human-carnivore conflicts (Carter

et al. 2012a; Inskip et al. 2014), the majority of studies to date have described conflicts or patterns

of attitudes towards carnivores, whereas fewer studies have focused on the underlying drivers and

impacts of conflict, or on conflict resolution and coexistence (Lescureux & Linnell 2010; Barua et

al. 2013; Can et al. 2014). In Europe, such knowledge is particularly important because recent

expansions of brown bear (Ursus arctos) populations have caused conflicts (Enserink & Vogel 2006;

Can et al. 2014), and illegal killing could undermine the recovery of bear populations (Ciucci &

Boitani 2008; Kaczensky et al. 2011). To understand coexistence, regions where humans and

carnivores have successfully co-occurred for a long time could provide particularly useful case

studies (Boitani 1995).

In this study, we aimed to elicit the social drivers of coexistence by taking a combined

approach of questionnaires and inductive semi-structured interviews to examine human-brown

bear coexistence. We focused on the foothills of the Carpathian Mountains in Transylvania,

Romania. Here, people and bears have co-occurred for extended periods of time, and the close

proximity of the forest to villages and farmland, as well as the reliance of people on forest

ecosystem services (e.g. firewood) provide ample opportunities for human-bear interactions.

Furthermore, traditional practices such as shepherding and bee keeping are potential areas of

conflict with bears (Zedrosser et al. 2001).


191

Our specific objectives were (i) to obtain a general overview of how social drivers affect

people’s perception of human-bear coexistence; and (ii) to generate a deeper understanding of the

mechanisms and causal factors underlying people’s perceptions of coexistence. One of our key

outcomes is the concept of coexistence pathways, which show how ongoing interactions between

elements of the ecological system and the social system shape the willingness of people to coexist

with carnivores. We outline specific coexistence pathways for our case study, and discuss their

potential general relevance for carnivore conservation. In addition, we discuss how the concept of

co-existence pathways could be extended to other species and locations by providing a general

framework for unpacking the social factors mediating human-carnivore co-existence.

Methods

Our research design was based on a sociological triangulation approach which combined

quantitative and qualitative methods to broaden understanding rather than to validate each other

(Olsen 2004). In particular, we used quantitative questionnaires to identify general patterns, and

inductive, semi-structured qualitative interviews to gain a deeper understanding of the social factors

mediating human-carnivore coexistence. We used an inductive approach to generate insights on

human-carnivore coexistence, rather than a deductive approach to test specific theories or

hypotheses (Pratt 2009). This approach was chosen to construct a holistic framework of

understanding that would account for a wide range of social factors that shape co-existence.

Indeed, inductive research has previously been useful in gathering detailed information for

understanding human-carnivore relationships (Inskip et al. 2014).

Questionnaires for an overview

We used questionnaires to identify general patterns in the social drivers underlying coexistence

(objective i), by assessing how combinations of predefined social factors influenced people’s

perception of current and future coexistence. Questionnaires are useful when the researchers know

what they are seeking, give the opportunity to sample many individuals and thereby obtain an

impression of average opinions, and simplify the comparison between respondents by enabling

quantitative analyses (Huntington 2000). We utilised these strengths by basing questions on social

factors we deemed relevant to our study area, including socio-demographic factors, interactions and

conflicts with bears, general attitudes, knowledge acquisition and culture, perceived benefits and

disadvantages of bears, and bear management (e.g. Kaczensky et al. 2004; Dickman 2010; Redpath

et al. 2012; Dickman et al. 2014). For all questions on perceptions we used a 5-point Likert scale

(from strongly disagree to strongly agree), whereas for all other questions we either used multiple

choice or yes/no answers (46 questions total; Supplementary Material Text S9.1). Questions that

could not be answered by the respondent were noted as ‘don’t know’.

We randomly selected 30 villages of the over 400 villages scattered throughout the 7441

km2 study area spanning the full range of biophysical and social conditions (Dorresteijn et al. 2014).

Chapter IX

192

The average number of inhabitants per village was 584 (min–max cc. 30–1900) (INS 2011). We

aimed to ask 7-10 persons per village for the questionnaires in all 30 villages, based on chance

encounters, and obtained 252 responses.

General patterns on how social drivers influenced people’s perception of coexistence were

analysed using hierarchical agglomerative cluster analysis (Wards clustering based on Euclidean

distances: agglomerative coefficient of 0.95) and principal component analyses (PCA). First, we

used a cluster analysis to identify groups of people that were similar in their perceptions of the

importance of different social factors (questions based on a 5-point Likert scale). We characterized

these groups based on their perceptions of coexistence, interactions and conflicts with bears, and

socio-demographic factors. Second, we used PCA to extract the main social drivers of coexistence

and related these to four groups previously obtained from the cluster analysis. We calculated four

separate PCAs for each of the following themes: (i) attitudes towards bears; (ii) cultural values and

bear knowledge acquisition; (iii) benefits and disadvantages related to bears; and (iv) bear

management (Supporting information Fig. S9.1-S9.4). For each theme, we plotted the first two axes

and overlayed them with the four groups from the cluster analysis. Twenty individuals missed one

question, and their responses were replaced with imputed values (i.e. the average of the total

sample pool). All questions that were answered with ‘don’t know’ where scored with a 3 (neutral

opinion about a statement). All statistical analysis were implemented in the ‘R’ environment (R

Core Team 2013).

Semi-structured interviews for in-depth analysis

In contrast to questionnaires, semi-structured interviews give the opportunity to go deeper into

certain topics by guiding participants into discussions but allowing the interview to follow the

thoughts of the participant (Huntington 2000). Therefore, interviews can generate insights into the

mechanisms and causal factors underlying people’s perception of human-bear coexistence

(objective ii), and can help interpret the patterns derived from questionnaires. Although the

interviews were guided to prompt discussions, we allowed for discussions on unanticipated themes.

The themes covered were the same as in the questionnaire (Supplementary Material Text S9.2). All

interviews were conducted by a local Romanian and Hungarian speaker, digitally recorded, and later

transcribed and translated into English.

We aimed to interview 3-4 people per village in a total of 20 villages, and selected the 10

villages with the highest and 10 with the lowest perceived human-bear conflicts as indicated by

previous research (Dorresteijn et al. 2014) to ensure a broad gradient in perceived conflicts, and

thus a broad range of responses on conflict and coexistence. We purposefully sampled for

shepherds, foresters and hunters to obtain a wide variety in human-bear interactions. In total, we

conducted 70 interviews.

We analysed the interview transcripts by first grouping all interview participants into three

groups regarding their perception of human-bear coexistence (positive, negative and neutral). For


193

ease of communication we only show the results here for people with positive or negative

perceptions. We applied qualitative content analysis by coding the interviews for themes using

‘NVivo 10’ (QSR International Pty ltd 2012). Coding was conducted in an exploratory way to

extract themes that related to those covered in the interview guide as well as to unprompted

content or emerging themes. To gain a deeper understanding of the mechanisms driving

coexistence, we combined themes in an emergent conceptual framework that included three

coexistence pathways. These coexistence pathways reflected how different elements of the social

and ecological systems interacted to create mechanisms that shaped human-bear coexistence

according to local people. Each coexistence pathway consisted of different themes, and in a second

round of coding, we categorized the statements of the participants to the different themes within

each pathway.

Results

Social drivers affecting perceptions, based on questionnaires

The cluster analysis revealed four distinct groups of respondents (Fig. 9.1), which differed in their

perception of coexistence, interactions with bears, and socio-demographic factors (Table 9.1). The

majority of people clustered in groups 1 and 2 and had a more positive perception of current and

future human-bear coexistence compared to groups 3 and 4. Respondents from group 1 had the

fewest interactions with bears, and respondents from groups 3 and 4 most frequently experienced

damage to crops and predation on livestock. The largest proportion of women was in group 1 and

the largest proportion of men was in group 2. Respondents in group 4 were oldest on average.

The identified groups revealed that attitudes towards bears, culture and knowledge

acquisition, perception of damages and threats, and opinion of management, may be important

social drivers influencing people’s perceptions of human-bear coexistence. The four groups of

respondents were characterized by their position along the main gradients of the four themes (see

PCAs in Fig. 9.1; full ordinations can be found in Supporting Information, Fig. S9.1-S9.4). Overall

patterns were more similar between groups 1 and 2, who had a more positive attitude towards

bears and between groups 3 and 4, who held a more negative attitude (Fig. 9.1). Similarly, groups 1

and 2 perceived bears to be less harmful than groups 3 and 4, while in contrast different beliefs

regarding the benefits provided by bears were not reflected in the grouping (Fig. 9.1). Also,

responses on the influence of culture, knowledge acquisition, and management had relatively little

influence on the grouping. However, group 4 ascribed low cultural importance to bears, and group

3 acquired most knowledge through experience (Fig. 9.1). Opinions on management most

importantly were characterized by the desire to be more actively involved, which was especially

high for group 2, whereas the overall satisfaction with bear management differed only slightly

between the groups (Fig. 9.1).

Chapter IX

194

Figure 9.1 Classification and ordination of the 252 questionnaire respondents according to their perceptions

of bear-related themes. The left panel shows the dendrogram and the four groups of people derived from

hierarchical agglomerative cluster analysis. The other panels show the principal components analyses of the

four different themes. The upper row shows the main gradients, whereas the lower four rows show the

loadings for each group on the different PCAs. The legs of the spider diagrams indicate each person

belonging to a given group and the circle indicates the standard deviation of the weighted average of each

group.


195

Table 9.1 Perception of current and future coexistence, socio-demographics, frequency of actual bear

encounters and conflicts, and average knowledge score, of the four different groups derived from the cluster

analysis based on the questionnaires. The number of people and the percentage in parentheses within a

certain group are given.

Mechanisms that shape perceptions, derived from interviews

Coexistence pathways

Similar to the questionnaires, the majority of respondents in the interviews had a positive

perception of human-bear coexistence (61%; n = 50), whereas 18 people (25%) had a negative

perception, and 10 people (14%) did not have a very strong opinion. Three coexistence pathways

emerged from the data and showed how people’s attitudes towards different aspects of bears, the

human community, and management support or oppose their willingness to live with bears (Fig.

9.2). Furthermore, the pathways related to bears and humans were affected through interactions

with the landscape (Fig. 9.2). However, the identified pathways do not predict perceptions of

coexistence, and each pathway can lead to both positive and negative perceptions of coexistence,

depending on the respondent. Rather, a pathway refers to the mechanisms that create a person’s

relationship with bears, and how that relationship is perceived depends on how the respondent

experiences those mechanisms.

Group 1

(n = 84)

Group 2

(n = 87)

Group 3

(n = 50)

Group 4

(n = 31)

Current Coexistence

Positive

Negative

70 (83)

14 (17)

82 (94)

4 (5)

24 (48)

25 (50)

18 (58)

13 (42)

Future Coexistence

Better

No change

Worse

9 (11)

35 (42)

23 (27)

20 (23)

44 (51)

13 (15)

2 (4)

15 (30)

29 (58)

2 (6)

7 (23)

15 (48)

Gender

Male (n = 181)

Female (n = 71)

46 (55)

38 (45)

78 (90)

8 (9)

34 (68)

16 (32)

23 (74)

8 (26)

Average age 47 46 47 62

Frequent bear

observations 7 (8) 22 (25) 19 (38) 9 (29)

Frequent damage to crops,

orchards and hives 6 (7) 5 (6) 21 (42) 6 (19)

Frequent attacks on livestock 1 (1) 2 (2) 8 (16) 2 (6)

Chapter IX

196

Figure 9.2 Conceptual framework showing three identified coexistence pathways based on qualitative

content analyses of 70 interviews. A coexistence pathway shows how ongoing interactions between elements

of the ecological system and the social system shape the willingness of people to coexist with bears.

Landscape-bear coexistence pathway

The landscape-bear coexistence pathway contained two components. The first component was

related to how people conceptualise interactions between bears and the surrounding landscape,

while the second component described people’s opinion on bear behaviour and ecology. The

pathways demonstrated that coexistence was supported by people’s understandings of bear

behaviour, and hindered by concerns about inadequate bear habitat, deforestation, and increasing

bear populations. The make-up of the surrounding landscape played an important role in people’s

perception of current and future coexistence (Table 9.2, row a). People with a positive perception

deemed forest size and food supply in the region sufficient, while people with a negative perception

deemed it insufficient. Deforestation and land-use change were major concerns of both groups

because they expected an increase in future conflicts with increasing disturbance to bear habitat

(Table 9.2, row b).

The ‘peaceful’ behaviour of bears and their relatively low damage compared to other

species (e.g. wolf) were character traits considered important to coexistence by both groups (Table

9.2, row c). The importance of perceptions of bear behaviour to coexistence was further expressed

through the local belief of the existence of two types of bears, namely the primarily vegetarian ‘ant-

eating bear’, and the ‘carnivorous bear’ (Table 9.2, row d). Coexistence with the ant-eating bear was

perceived to be unproblematic, whereas the carnivorous bear was perceived as a dangerous animal.

Despite the view of bears being relatively harmless, the perceived increase in the bear population


197

was considered a major problem to coexistence (Table 9.2, row e). Several people even mentioned

that bears were only present recently and that they either came down from the mountains or had

been brought to the area from overpopulated areas or for hunting purposes.

Table 9.2 Characterizing quotes for the themes within the landscape-bear coexistence pathway. The

respondents were grouped into two groups based on their perception of coexistence (positive or negative).

The number of people mentioning a given theme is reported and the percentage within each group is given in

parentheses. The ‘P’ or ‘N’ behind each quote indicates whether the person had a positive or negative

perception of coexistence.

Landscape-bear coexistence pathway

Perception of coexistence

Positive

(n = 43)

Negative

(n = 18) Characterizing quotes

a. Bear resources 17(40) 6(33) So long as forests remain I don’t think something bad can

happen, I don’t think the bear will come down in the village

(ALM4; P);

As long as it has food in the forest it won't eat up the people's

potatoes or corn. Then it has no reason to come into the village

(SAC2; P);

We have only a few, small forests here! There's no place for the

bear to stay there (BLA3; N).

b. Land-use

change

17(40) 4(22) The bear typically stays far from humans. Only when it is being

attacked or when it has cubs they are dangerous (ALM1; P);

I'm thinking about deforestation. You destroy the bears’

habitat. It needs to adapt as well. It can't hide anymore; it gets

more and more in contact with humans, its hunting area

disappears and that becomes a problem. This is how the bear

may become a problem! (MAL1; N).

c. Bear behaviour 21(72) 14(78) The bear is not an animal that attacks without being provoked

(BLA1; P);

The shepherd was always saying: ‘Hey, the bear is coming, the

bear is coming!’ [I answered]: ‘Let it come, man! Let it come’

When it comes, it takes [one/ a few]... It's not like the wolf! A

wolf, once it jumps into a compound, it kills 4-5-10 sheep, and

then it takes one and leaves! But the bear takes one under its

arm and leaves (DEA1; N).

d. The anteater

vs. the carnivore

12(28) 5(28) There are only ant-eating bears. They don't attack the sheep

(MAL2; P);

Doesn't matter what point of view we take on them: As long as

they eat plants, there's no problem with them, but once it gets

to taste meat, it'll get aggressive (SAC1; N).

e. Bear

population

25(58) 12(67) No, there aren't 'urgent' problems – it's just that they appear

more and more often! (VAD3; P);

Because there are more of them, it attacks animals and man

more often! (SAC4; N);

We have knowledge about the fact that bears have been

brought here. There haven't been that many bears in the past,

by far (ALE1; P);

It has been brought here. And then it reproduced (VAL3; N).

Chapter IX

198

Landscape-human coexistence pathway

The landscape-human coexistence pathway also contained two components. The first component

described the relationships between people and the landscape (including interactions with bears),

while the second component related to people’s values. Experiencing bears and nature, and the

values ascribed to them, were key in shaping people’s perceptions of coexistence. The perception

of coexistence was especially positive for people that had positive interactions with bears, while

negative perceptions were related to higher livestock predation and the levels of perceived damage

and danger by bears (Fig. 9.3a). Interestingly, such damage beliefs appeared to have more influence

on coexistence than actual conflict (Fig. 9.3a). In contrast, crop damage and indirect knowledge of

conflicts had little effect on the perception of coexistence (Fig. 9.3a). Moreover, preventing

conflicts by adjusting human behaviour or using traditional shepherding techniques was seen as an

integral component of coexistence by both groups and was mentioned in 62% and 23% of the

cases respectively.

Family, community members, and education all played a role in acquiring knowledge on

how to live with bears, but people with a positive perception mentioned more often that they

learned through experience (Fig. 9.3b). Attitudes towards bears were especially positive for people

with a positive perception of coexistence (88% positive attitude; 0% negative attitude), compared

to people with a negative perception (19% positive attitude; 28% negative attitude). Non-use

values, such as cultural, existence, and historical values, were more often associated with bears than

use values by both groups, although these values were more prominent among people with a

positive perception (Fig. 9.3c). The genuine and continuous relations between people and nature,

where they experience and value nature, seemed to be more important to support coexistence than

financial incentives or other use-values: ‘Well, for us the bear is like our neighbour. (…) like the

neighbour at home. There's no difference between my neighbour and the bear that comes every

second evening. We're somehow used to it. (VID4)’; ‘When you love nature I don't think that you

need a lesson, necessarily, to live with them [bears] (CIN3)’; You don’t want to see a dead forest

[without animals], right? It's different when you see a bird on a tree, a bear, a deer. ‘That makes the

difference! You see life in nature! Dead nature is like a village without inhabitants (LAS4)’. Fear was

one of the emergent themes but only mentioned by several people (afraid: n = 6; not afraid n= 9),

without large differences between the two groups.


199

Figure 9.3 Human-bear interactions, bear knowledge acquisition, and values ascribed to bears between two

groups of respondents – those with a generally positive perception (n = 43) of human bear-coexistence

versus those with a generally negative perception (n = 18). a) The difference between actual conflicts, bear

interactions and perceived damage risks. The actual conflicts and bear observations were derived from direct

questions asked at the beginning of the interview (see S9.2), while perceived damage risk, other direct

interactions, and indirect knowledge of conflicts was interpreted from the interviews. b) Sources where

people acquire knowledge on how to live with bears. c) Values ascribed to bears. The importance of cultural

values was a question in the interview guide (see S9.2) whereas all other types of values emerged from the

interviews.

Management coexistence pathway

The management-coexistence pathway was related to the institutions managing bears and how

people perceived these institutions. The majority of people were dissatisfied with current bear

management mainly due to a lack of concern for their problems with bears, disinterest from the

management authorities, distrust towards them, lack of conflict mitigation, and perceived bad

management (Table 9.3). The only 12 people that were satisfied with management believed that the

Chapter IX

200

provision of supplemental or divisionary feeding was enough to prevent conflict (Table 9.3, row c).

However, the majority of people believed that authorities and hunting organizations failed to

mitigate conflict by not feeding the bears (enough), by not controlling the bear population, and by

not providing compensation after damage (Table 9.3, rows i-k). Furthermore, six people expressed

concerns about trophy hunting, indicating that common people carried the burden of living with

bears so that hunting organizations could earn money from trophy hunting (Table 9.3, row l).

Although the law and protection status was perceived to aid coexistence by people with a positive

perception of coexistence, people with a negative perception viewed it as an artificial tool that they

could not influence or even felt disempowered by, because they were not able to take care of

problem animals themselves (Table 9.3, row g and m).

Table 9.3 Characterizing quotes for the themes within the management coexistence pathway. The

respondents were first grouped into two groups based on their perception of coexistence (positive or

negative), and second, based on their opinion whether they were satisfied with current bear management. The

number of people mentioning a given theme is reported and the percentage within each group is given in

parentheses. The ‘P’ or ‘N’ behind each quote indicates whether the person had a positive or negative

perception of coexistence.

Management coexistence pathway

Perception of coexistence

Positive

(n = 11)

Negative

(n = 1) Characterizing quotes

Satisfied with current bear management

a. Lack of

concern and

distrust

b. Conflict

mitigation:

compensation

1(9) Yes, there is the principle when an animal caused damage,

the person affected will be compensated. (…)There is this

principle of compensation at local level (BLA1;P).

c. Conflict

mitigation:

feeding

9(82) 1(100) Yes, they feed them – the ‘paznic de vanatoare’. This has

been done in the past as well, when they brought them dead

animals in a big open pit (CIN1; P).

e. Management:

control bear

population

1(9) I'm against their killing and wiping them out, but I think that

their population growth should be controlled a bit, as well as

their spreading (CIN3; P).

f. Management:

legal and illegal

hunting

6(55) Every year we shoot one bear which is pretty ok. That's the

money of the hunting association. They pay something to

the state as well. Don't know if they pay something to the

local community (VAN2; P).

g. Law and

disempowerment

5(45) They'd almost disappeared and [you can shoot them today]

only with a special permit where the cause big damage. The

wolf, the wild cat, the bear – they're all protected (GRA1; P).


201

Positive

(n = 21)

Negative

(n = 13)

Dissatisfied with current bear management

h. Lack of

concern and

distrust

11(52) 3(23) But they don't do anything. Nobody cares for nothing!

Nothing. I've been to the mayor and he said: ‘Should I come

and guard them instead of you?’(VID4; P).

A bear values four human lives. Yes! Once it has killed four

humans, you're allowed to shoot it. Therefore, it is worth

four human lives. (VAL3; N).

i. Conflict

mitigation:

compensation

5(24) 5(38) The one who has the damage, stays with the damage. Only

when the damage is bigger. Only then. Otherwise nobody

cares/makes an effort (ALM1; P);

I guard the fields that I have paid for, otherwise the wild

animals destroy the harvest and nobody pays me the damage.

Nobody pays for it! Because we don't have such a big output

(...).We don't have a big production but we're living of it. The

damages aren't big for them [authorities], but we have to live

from this (GHI1; N).

j. Conflict

mitigation:

feeding

8(38) 4(31) When there are many bears, they should guarantee them feed

so they don't come and attack…I2: They don't guarantee

them the feed, that's why they attack! (CRI2; P).

They must be fed as well! It's not enough to let them free in

the forest, but to feed them as well. The forest authority has

fed them in the past! (PRO3; N)

k. Management:

control bear

population

3(14) 3(23) Let the bear population grow but keep it under control

(AGA1; P).

The population management is a bit out of control. I don't

think there is a true population management. (…) But the

true control was in the past the fact that the bears were kept

under a certain number/density in a certain area (MAL1; N)

l. Management:

legal and illegal

hunting

13(62) 8(62) For the locals it's just disadvantages, cause the hunters are

not from this region so that we can say: ‘We keep the

earnings from the hunting!’ In case they shoot a bear. The

problem is that those who shoot the bears are not locals and

the truth is that, as far as I see, they don't even care for

compensating for losses that locals might have (SAC3; P);

The management of bears by the state is not working

properly. This could be improved – but as long as a hunter

may come to shoot a bear for a certain amount of Euros,

without caring that this causes to others a double amount of

damage or even a human's life... We'd need to think about all

this! Then bears and humans would get along well (MAD1;

N).

m. Law and

disempowerment

8(38) 5(38) How should I say: They could live together. Cause you can't

do them anything! You can't go in the forest and shoot them.

What can you do else, but live next to them (BLA2; P);

A human life is worth more than a bear's, Sir. Protecting

bears is good, but it's somehow overprotected, Sir (MIH1;

N).

Chapter IX

202

Discussion

Our methodological approach provided two different perspectives on the social drivers to human-

bear coexistence. First, questionnaires revealed general patterns of the most important drivers

within groups of people. Grouping people with similar perceptions may be beneficial for the design

of more specific conservation programs targeted at different groups or societal concerns. Second,

semi-structured interviews identified coexistence pathways highlighting the causal mechanisms

driving people’s willingness to coexist with bears. Our research showed that perceptions of

coexistence were negatively influenced by negative attitudes, past negative interactions with bears,

perceived risks of damage, and respondents’ age, which is in line with previous studies (e.g.

Naughton-Treves et al. 2003; Kaczensky et al. 2004; Carter et al. 2012a). Nevertheless, other factors

such as culture and management were also found to be important.

Our study showed that human-carnivore coexistence is possible, and we identified three

major pathways consisting of different social drivers that influenced human-carnivore coexistence

from the perspective of the rural population. These pathways serve as a framework for identifying

the drivers that shape co-existence and how they interact. Identifying such drivers is essential in

shaping management approaches, but also to linking the drivers to broader theoretical and research

areas that help to understand how these factors are shaped. Thus the concept of coexistence

pathways and the research approach to elicit them offers a conceptual framework to assist in

facilitating human-carnivore facilitation worldwide.

Factors mediating coexistence

The two landscape-mediated coexistence pathways showed that conservation management geared

towards facilitating coexistence needs to address the beliefs shaping people’s tolerance towards

carnivores, and maintain or re-establish people’s connections to carnivores and nature. Similar to

other regions (Lescureux et al. 2011), direct interaction with bears and experiential knowledge

acquisition was particularly important in shaping perceptions and beliefs about bears (e.g.

behaviour, anteater vs. carnivore, population size).

The importance of direct interactions was further emphasized through reduced tolerance

of people that experienced livestock predation, while positive direct interactions were distinctively

higher among people with a positive perception of coexistence. Nevertheless, beliefs about bear-

related conflicts and damage risk more strongly determined people’s perception of coexistence than

the actual negative experiences, which has also been observed elsewhere (Kaczensky et al. 2004;

Carter et al. 2012a). More importantly, a stronger impact of affective risk perceptions compared to

cognitive risk perceptions can influence human behaviour and has been associated to motivations

of tiger killing (Inskip et al. 2014). To fully comprehend human-carnivore coexistence therefore

requires an understanding of the links between beliefs and risk perceptions.

The landscape-human coexistence pathway also showed that key social drivers for people’s

perceptions of human-bear coexistence were landscape-mediated attitudes and non-use values.


203

These were constructed through the existence of genuine links between people and nature, where

people valued their surrounding landscape and considered bears as part of their natural heritage.

Thus, human-carnivore coexistence may be influenced by factors such as sense of place, and

understanding which specific dimensions of sense of place affect human-carnivore coexistence

could be an important domain for further research (Williams & Stewart 1998). Such values may be

more important drivers of people’s perceptions towards carnivores than damage risks (Lagendijk &

Gusset 2008; Glikman et al. 2012; Dickman et al. 2014). Tolerance may have been further

facilitated by low fear levels, most likely resulting from sharing the landscape with bears (Roskaft et

al. 2003; Majić et al. 2011). Moreover, in intact social-ecological systems cultural tolerance to

carnivores is not uncommon (Lagendijk & Gusset 2008) and can reduce carnivore extinction risk

(Karanth et al. 2010). Such a genuine link to nature is perhaps facilitated by the continuous

coexistence with large carnivores over long periods of time. Such long-term coexistence most likely

also influenced human behaviour to avoid conflict with bears and prevent livestock predation by

using traditional livestock husbandry techniques. Thus, the continuous coexistence with large

carnivores presumably shaped the emotional component of human culture to accept and adapt to

human-bear coexistence (Glikman et al. 2012). While a history of continuous coexistence cannot be

re-created in places where carnivores were once extirpated, a key conservation challenge for

settings without a continuous history of human-carnivore coexistence is to (re-)connect people to

nature and carnivores.

The management coexistence pathway underlined the perceived gaps in current

management and showed that perceptions of (mis-)management could become a major obstacle to

coexistence. Consistent with the varying responses to management in the questionnaire, strong

negative opinions about current management did not necessarily lead to a negative perception of

coexistence. Nevertheless, people’s negative opinions on various aspect of current bear

management and the feeling of being treated unfairly have the potential to erode the build-up

tolerance towards carnivores through human-human conflicts. For example, a perception of

inadequate governance can compel people to retaliatory killing (Treves et al. 2002), and poaching in

Greece is partly motivated by a desire to defy the authorities (Bell et al. 2007). The approach of

seeking coexistence pathways was particularly useful to detect where people perceived mitigation as

necessary, which strategies were culturally accepted, and whether current efforts were satisfactory.

For example, the beliefs and concerns around an increasing bear population and its control should

receive high priority because increasing carnivore populations, rumours about ‘secret’ carnivore

introductions, and inequality over benefits and disadvantages, could reduce people’s tolerance

towards carnivores (Skogen et al. 2008).

Chapter IX

204

Managing coexistence

To avoid such disagreements between management bodies and local stakeholders, management

may facilitate coexistence through participation and education. Distrust and the feeling of

disempowerment can be reduced by including people in carnivore management through

participatory processes or active management (Treves et al. 2006), and conflicts could be reduced

through a holistic set of mitigation strategies geared towards the needs of both humans and bears

(Can et al. 2014). Although the interviewed respondents did not express the wish to be involved in

bear management, several respondents to the questionnaires showed this aspiration. This potential

for co-management of different local stakeholders may increase or maintain tolerance through

shared responsibilities, accepted management and good relationships between local people and

management bodies (Treves et al. 2006; Lagendijk & Gusset 2008).

Such participation or collaboration could also provide an education component or

function. Under the landscape-mediated pathways, experience and education were mentioned as

the most important sources to acquire knowledge about bears, and conservation could use these

tools to address people’s beliefs and concerns to facilitate coexistence. For example, by fostering

positive interactions with bears, people may feel less threatened and their tolerance towards bears

may increase (Majić et al. 2011). Education can help to foster tolerance towards large carnivores by

targeting specific (local) beliefs (Zinn et al. 2008).

Finally, direct actions such as well-regulated and managed options for local people to react

against certain ‘carnivorous’ problem animals could increase empowerment of people (Lescureux &

Linnell 2010; Majić et al. 2011). Yet, non-lethal interventions are still more effective in reducing

livestock predation rates (Bergstrom et al. 2014), and socially accepted methods like divisionary

feeding need more attention. The anger around compensation payments may be harder to resolve

when it is governed by perceived ‘weak’ institutions (Ferraro & Kiss 2002). Although compensation

payments or other financial benefits can aid conservation (Dickman et al. 2011), they often do not

improve tolerance (Naughton-Treves et al. 2003; Hazzah et al. 2009) or are distributed unequally

among society (Hemson et al. 2009). Therefore it could be worthwhile to explore other alternatives

for social-ecological systems to manage human-carnivore conflicts such as bottom-up community

organized payments like contributions to a local livestock insurance program (Mishra et al. 2003).


Facilitating human-carnivore coexistence is a conservation goal worldwide. By highlighting the

existence of coexistence pathways, and then unpacking these pathways, it is possible to derive

meaningful recommendations for approaches to manage coexistence. In our case study, the content

of conservation programs should be designed to target these mechanisms in a way that

meaningfully engages with people’s coexistence perceptions, both in terms of content and how it is

delivered. In particular, we advocate for a more collaborative, participatory approach to carnivore

management in this area. The approach should foster people’s connection to their landscape,


205

provide education on bears and management approaches, and provide transparency around

management.

More broadly, we believe that our approach of combining questionnaires for large scale

patterns with the concept of coexistence pathways could be extended to regions worldwide. In

particular, the inductive approach of generating an understanding of coexistence pathways could be

used to understand human-carnivore coexistence for different regions and species. Whether the

specific pathways and their outcomes are transferable to other regions remains yet to be tested in

future studies. However, by constructing co-existence pathways through inductive research, a

broad range of factors that influence co-existence can be identified, and how these factors interact

with each other can be understood. We therefore encourage scientists to use interdisciplinary

research to obtain a comprehensive understanding of social drivers to human-carnivore

coexistence, and thus to work towards holistic conservation efforts. Through the framework

concept of coexistence pathways, and a comprehensive analysis of such pathways, landscape

managers can identify factors that facilitate or hinder coexistence, and thus target interventions

accordingly.

Acknowledgements

We are grateful to Hans Hedrich for conducting and translating all the interviews and we thank all

respondents for their participation. The survey procedure was cleared by the ethics committee of

Leuphana University Lueneburg. The project was funded by a grant of the International

Association for Bear Research and Management to ID and a Sofja Kovalevskaja Award by the

Alexander von Humboldt Foundation to JF.

Chapter IX

206

Supplementary Material for Chapter IX

Text S9.1. The questionnaire used for 252 people (original in Romanian).

Please indicate how often: 1. You see a bear

a) never b) rarely c) several times a year d) often

2. A bear damages your fields/orchards/beehives a) never b) rarely c) several times a year d) often e) not applicable

3. A bear attacks your animals a) never b) rarely c) several times a year d) often e) not applicable

The following questions will ask if you use any protective measures against bears Please indicate:

4. Do you use protective measures to prevent damage to fields/orchards/beehives a) no b) occasional c) yes d) not applicable

5. Do you use protective measures to prevent bear attacks on your animals

a) no b) occasional c) yes d) not applicable

6. I go into areas where bears are present a) no b) occasional c) yes

7. In my daily life I try to prevent getting into conflict with bears

a) no b) occasional c) yes The following questions are general questions about bears Please answer the following statements to your best knowledge:

8. Bears mainly feed on meat a) yes b) no

9. Female bears have young every year a) yes b) no

10. Most bears weigh more than 150 kg a) yes b) no

11. Bear cubs leave the mother in their first year of their life a) yes b) no

12. Bears are protected animals in Romania a) yes b) no The following questions will ask about how you achieve knowledge about bears Please indicate the extent to which you disagree or agree with the following statements: 1 = strongly disagree; 2= disagree; 3 = neutral; 4 = agree; 5 = strongly agree

13. My parents told me how to live in a landscape with bears 1 2 3 4 5

14. At school we learned about bears and how to live with them 1 2 3 4 5

15. Local authorities provide information on how to live with bears 1 2 3 4 5

16. Local NGO’s provide information on how to live with bears 1 2 3 4 5


207

17. What I know about bears, I learned from experience 1 2 3 4 5

The following questions will ask about bears in your culture Please indicate the extent to which you disagree or agree with the following statements

18. Bears are important in our culture 1 2 3 4 5

19. Hunting bears is important in our culture 1 2 3 4 5

20. I grew up hearing stories about bears 1 2 3 4 5

21. In the stories I know, bears mainly have a positive character 1 2 3 4 5

22. It is important for our culture that bears persist in the landscape 1 2 3 4 5

The following questions are regarding your feelings towards bears in general Please indicate the extent to which you disagree or agree with the following statements

23. I generally like bears 1 2 3 4 5

24. It is bad to have bears in Transylvania 1 2 3 4 5

25. Bears should remain part of our landscape in the future 1 2 3 4 5

26. I am afraid to meet a bear 1 2 3 4 5

27. Bears do not have the same rights as humans to exist in the landscape 1 2 3 4 5

The following questions are regarding the usefulness of bears in the landscape Please indicate the extent to which you disagree or agree with the following statements

28. Bears have a negative impact on hunting opportunities 1 2 3 4 5

29. In areas where there are bears and sheep, bears kill a lot of sheep 1 2 3 4 5

30. Bears damage a lot of orchards and fields 1 2 3 4 5

31. Bears are dangerous to humans 1 2 3 4 5

32. Bears increase the value of a hunting area 1 2 3 4 5

33. Having bears increases tourism in the area 1 2 3 4 5

34. Bears keep the forest clean of dead and sick animals 1 2 3 4 5

35. Bears keep nature in balance 1 2 3 4 5

Chapter IX

208

The following questions are regarding the management of bears Please indicate the extent to which you disagree or agree with the following statements

36. Bears should be completely protected 1 2 3 4 5

37. Local authorities put in enough effort to prevent damage by bears 1 2 3 4 5

38. Hunting associations put in enough effort to prevent damage by bears 1 2 3 4 5

39. I would like to be involved in the management of bears in the area 1 2 3 4 5

40. We receive compensation for damage by bears 1 2 3 4 5

41. Trophy hunting benefits the entire community 1 2 3 4 5

42. Hunting bears should be possible to everybody in the community 1 2 3 4 5

The following questions are about your opinion on how humans and bears share the landscape

43. In your opinion, how do bears and humans live together in this region a. Peacefully without conflicts b. Relatively peacefully with tolerance for occasional conflicts c. Relatively unpeacefully due to occasional conflicts d. Unpeacefully due to escalating conflicts

44. In your opinion, how do you see the relationship between humans and bears in the future?

a. Better b. Worse c. No change

45. Explain shortly your answer under question 44

___________________________________________________________________ ___________________________________________________________________ ___________________________________________________________________

46. Would you like to add any additional information about bears?

___________________________________________________________________ ___________________________________________________________________ ___________________________________________________________________

The following questions are with respect to you IX. Gender: a) M b) F X. Age:______years

XI. Profession XII. Ethnicity: a) Romanian b) Hungarian c) Roma d)Saxon e) Other, namely:

XIII. Where did you grow up: a) this region b) a different region, but with bears present c) a different region without any bears present



209

Text S9.2. The interview guide used for semi-structured interviews with 70 people (original in

Romanian).

1. The following questions will ask about your experience with bears

Please indicate how often:

1.1 You see a bear

a) never b) rarely c) several times a year d) often

1.2 A bear damages your fields and/or orchards, and/or beehives

a) never b) rarely c) several times a year d) often e) not applicable (no

fields/orhards/beehives)

1.3 A bear attacks your animals

a) never b) rarely c) several times a year d) often e) not applicable (no

animals)

1.4 Which other animals cause problems in the village? Do you think these problems are worse

than the problems caused by bears?

___________________________________________________________________

___________________________________________________________________

___________________________________________________________________

2. (Traditional ecological) knowledge on how to live with bears

Romania is a special country regarding bears because it has one of the largest populations in

Europe. Furthermore, Transylvania is especially unique, because people and bears live alongside

each other and share the same landscapes. This is very different from Western Europe. There are

few bears in Western Europe, and in many places they have been hunted to extinction. However, at

the moment in many places bears are returning to Western Europe. Therefore, we would like to

know and learn more on how you manage to live together with bears in Transylvania.

2.1 How do you manage to live together with bears?

2.2 Where and how do you learn to share a landscape with bears?

3. Which factors could disturb human-bear relationships

The Transylvanian country side has changed rapidly over the last years. For example, tourism from

foreign countries has increased and the architecture of the houses has changed. In your opinion,

are there certain changes that have changed or might change the relationships between humans and

bears?

3.1 Do you think the way people and bears live together now is different from the past?

3.2 Do you think people and bears can live together in this region in the future?

3.3 What do you think are the major factors that could change the way people currently live

with bears?

Chapter IX

210

4. Cultural values and attitudes towards bears

In some countries bears are important for their culture and there are stories and believes around

bears. For example, Sleeping Bear Dunes, a huge sand dune in America, is named after a Native

American legend. A female bear and her cub swam across a big lake (Lake Michigan). Exhausted

from their journey, the bears rested on the shoreline and fell sound asleep. Over the years, the sand

covered them up, creating a huge sand dune. Or for example, in Finland and Russia the bear is the

national animal. In your opinion, does the bear play an important role in your culture?

4.1 Are bears an important part of your culture?

4.2 Can you remember any particular stories that include bears? Which one and what kind of

character did the bear have?

Germany and Switzerland have had similar recent experiences with bears. In both countries one

bear came back and lived in the country for a while. However, in both countries they decided to

shoot the bear as it was classified as a problem bear and a threat to human safety. How do you feel

about this approach?

4.3 What would you do in this case? What are your feelings in general towards bears?

4.4 Do you think it is important that bears persist in the Transylvanian landscape in the future?

5. Management of bears

Not everybody in a country, region or village is affected in the same way by sharing the landscape

with bears. Do you think that there are any benefits or disadvantages of sharing the landscape with

bears?

5.1 Are these benefits or disadvantages equally distributed between community members?

5.2 Do local authorities/hunting associations help to prevent damage caused by bears? Do

they take care to close the difference in the distributions of benefits/disadvantages

indicated in the previous question?

People living in countries that lack large carnivores such as bears are often very impressed by the

presence of these animals. Therefore, many countries have developed tourism based on large

carnivores. This includes guiding tourists around the landscape with the chance to see and

photograph carnivores, or tourists pay for a license to hunt for carnivores and they can take home

the fur and parts of the skeleton (trophy hunting). Are there any of these examples present in your

village?

5.3a If yes, are they managed well and does the community benefit from this? Who benefits?

5.3b If no, do you see a future for tourism based on bears in this region?

The following questions are with respect to you

I. Gender: a) M b) F

II. Age:

III. Profession

IV. Ethnicity: a) Romanian b) Hungarian c) Roma d)Saxon

V. Where did you grow up: a) this region b) a different region, but with bears present

c) a different region without any bears present



211

Figure S9.1 Principal Component Analyses of all questions reflecting the participant’s attitudes towards

bears. The numbers reflect the four different groups derived from the hierarchical agglomerative cluster

analysis. The ‘QX’ indicates the question from the questionnaire in Supporting Information Text S9.1. The

PCA was based on the following six questions: 23-26, 36, 42.

Figure S9.2 Principal Components Analysis of all questions reflecting the participant’s perception of cultural

values of bears and knowledge acquisation about bears. The numbers reflect the four different groups

derived from the hierarchical agglomerative cluster analysis. The ‘QX’ indicates the question from the

questionnaire in Supporting Information Text S9.1. The PCA was based on the following six questions: 13-

22.

Chapter IX

212

Figure S9.3 Principal Components Analysis of all questions reflecting the participant’s perception of bear-

related benefits and disadvantages. The numbers reflect the four different groups derived from the

hierarchical agglomerative cluster analysis. The ‘QX’ indicates the question from the questionnaire in

Supporting Information Text S9.1. The PCA was based on the following six questions: 28-35.

Figure S9.4 Principal Components Analysis of all questions reflecting the participant’s perception of current

bear management. The numbers reflect the four different groups derived from the hierarchical agglomerative

cluster analysis. The ‘QX’ indicates the question from the questionnaire in Supporting Information Text S9.1.

The PCA was based on the following six questions: 37-41.

213

References

214

References

215

References

Akeroyd, J. R., and N. Page. 2007. The Saxon villages of southern Transylvania: conserving

biodiversity in a historic landscape. In: D. Gafta, and J. Akeroyd (eds). Nature

conservation: concepts and practice, Springer, New York, pp 199-210.

Albuquerque, F. S., M. J. T. Assunção-Albuquerque, L. Cayuela, R. Zamora, and B. M. Benito.

2013. European bird distribution is “well” represented by Special Protected Areas: mission

accomplished? Biological Conservation 159:45-50.

Alder, D., and S. Marsden. 2010. Characteristics of feeding‐site selection by breeding Green

Woodpeckers Picus viridis in a UK agricultural landscape. Bird Study 57:100-107.

Andrén, H. 1994. Effects of habitat fragmentation on birds and mammals in landscapes with

different proportions of suitable habitat: a review. Oikos 71:355-366.

Angelstam, P., L. Boresjö-Bronge, G. Mikusiński, U. Sporrong, and A. Wästfelt. 2003. Assessing

village authenticity with satellite images: a method to identify intact cultural landscapes in

Europe. Ambio 32:594-604.

Angelstam, P., and G. Mikusinski. 1994. Woodpecker assemblages in natural and managed boreal

and hemiboreal forest: a review. Annales Zoologici Fennici 31:157-172.

Apostolopoulou, E., and J. D. Pantis. 2009. Conceptual gaps in the national strategy for the

implementation of the European Natura 2000 conservation policy in Greece. Biological

Conservation 142:221-237.

Arion, F. H., I. Mures ̧an, and L. Balint. 2011. Compiling a visitors profile in Tarnava Mare area.

Bulletin of University of Agricultural Sciences and Veterinary Medicine Cluj-Napoca.

Horticulture 68:16-22.

Arnold, T. W. 2010. Uninformative parameters and model selection using Akaike's information

criterion. The Journal of wildlife management 74:1175-1178.

Athreya, V., M. Odden, J. D. C. Linnell, J. Krishnaswamy, and U. Karanth. 2013. Big cats in our

backyards: persistence of large carnivores in a human dominated landscape in India. PLoS

ONE 8:e57872.

Atkinson, P. W., R. J. Fuller, and J. A. Vickery. 2002. Large-scale patterns of summer and winter

bird distribution in relation to farmland type in England and Wales. Ecography 25:466-

480.

Baker, P. J., and S. Harris. 2006. Does culling reduce fox (Vulpes vulpes) density in commercial

forests in Wales, UK? European Journal of Wildlife Research 52:99-108.

Baldi, A., and P. Batary. 2011. Spatial heterogeneity and farmland birds: different perspectives in

Western and Eastern Europe. Ibis 153:875-876.

Banerjee, K., Y. V. Jhala, K. S. Chauhan, and C. V. Dave. 2013. Living with lions: the economics of

coexistence in the Gir Forests, India. PLoS ONE 8:e49457.

Barbaro, L., and I. Van Halder. 2009. Linking bird, carabid beetle and butterfly life-history traits to

habitat fragmentation in mosaic landscapes. Ecography 32:321-333.

Barber-Meyer, S. M., L. D. Mech, and P. J. White. 2008. Elk calf survival and mortality following

wolf restoration to Yellowstone National Park. Wildlife Monographs 169:1-30.

Barea-Azcon, J. M., E. Virgos, E. Ballesteros-Duperon, M. Moleon, and M. Chirosa. 2007.

Surveying carnivores at large spatial scales: a comparison of four broad-applied methods.

Biodiversity and Conservation 16:1213-1230.

Barnosky, A. D., P. L. Koch, R. S. Feranec, S. L. Wing, and A. B. Shabel. 2004. Assessing the

causes of Late Pleistocene extinctions on the continents. Science 306:70-75.

Barnosky, A. D., N. Matzke, S. Tomiya, G. O. Wogan, B. Swartz, T. B. Quental, C. Marshall, J. L.

McGuire, E. L. Lindsey, and K. C. Maguire. 2011. Has the Earth's sixth mass extinction

already arrived? Nature 471:51-57.

References

216

Barrientos, R. 2010. Retention of native vegetation within the plantation matrix improves its

conservation value for a generalist woodpecker. Forest Ecology and Management 260:595-

602.

Barua, M., S. A. Bhagwat, and S. Jadhav. 2013. The hidden dimensions of human-wildlife conflict:

health impacts, opportunity and transaction costs. Biological Conservation 157:309-316.

Batary, P., B. Andras, D. Kleijn, and T. Tscharntke. 2011a. Landscape-moderated biodiversity

effects of agri-environmental management: a meta-analysis. Proceedings of the Royal

Society B: Biological Sciences 278:1894-1902.

Batary, P., J. Fischer, A. Baldi, T. O. Crist, and T. Tscharntke. 2011b. Does habitat heterogeneity

increase farmland biodiversity? Frontiers in Ecology and the Environment 9:152-153.

Batáry, P., T. Matthiesen, and T. Tscharntke. 2010. Landscape-moderated importance of hedges in

conserving farmland bird diversity of organic vs. conventional croplands and grasslands.

Biological Conservation 143:2020-2027.

Bates, D., M. Maechler, B. Bolker, and S. Walker. 2014. lme4: Linear mixed-effects models using

Eigen and S4. R package version 1.0-4.

Bauhus, J., K. Puettmann, and C. Messier. 2009. Silviculture for old-growth attributes. Forest

Ecology and Management 258:525-537.

Bell, S., K. Hampshire, and S. Topalidou. 2007. The political culture of poaching: a case study from

northern Greece. Biodiversity and Conservation 16:399-418.

Benton, T. G., J. A. Vickery, and J. D. Wilson. 2003. Farmland biodiversity: is habitat heterogeneity

the key? Trends in Ecology & Evolution 18:182-188.

Berg, Å., and T. Gustafson. 2007. Meadow management and occurrence of corncrake (Crex crex).

Agriculture, Ecosystems & Environment 120:139-144.

Berg, Å., and M. Hiron. 2012. Occurrence of Corncrakes Crex crex in mosaic farmland landscapes

in south-central Sweden: effects of habitat and landscape structure. Bird Conservation

International 22:234-245.

Berger, J., P. B. Stacey, L. Bellis, and M. P. Johnson. 2001. A mammalian predator-prey imbalance:

grizzly bear and wolf extinction affect avian neotropical migrants. Ecological Applications

11:947-960.

Berger, K. M., E. M. Gese, and J. Berger. 2008. Indirect effects and traditional trophic cascades: a

test involving wolves, coyotes, and pronghorn. Ecology 89:818-828.

Bergmeier, E., J. Petermann, and E. Schroeder. 2010. Geobotanical survey of wood-pasture

habitats in Europe: diversity, threats and conservation. Biodiversity and Conservation

19:2995-3014.

Bergstrom, B. J., L. C. Arias, A. D. Davidson, A. W. Ferguson, L. A. Randa, and S. R. Sheffield.

2014. License to kill: reforming federal wildlife control to restore biodiversity and

ecosystem function. Conservation Letters 7:131-142.

Beschta, R. L., and W. J. Ripple. 2009. Large predators and trophic cascades in terrestrial

ecosystems of the western United States. Biological Conservation 142:2401-2414.

Bibby, C. J., N. D. Burgess, D. A. Hill, and S. Mustoe 2000. Bird census techniques. Academic

Press, London, 302 pp.

Bignal, E. M., and D. I. McCracken. 2000. The nature conservation value of European traditional

farming systems. Environmental Reviews 8:149-171.

Blom, S. 2012. Hur påverkas natur- och kulturvärden av en striktare betesmarksdefinition?,

Jordbruksverket rapport 20, available at http://www2.jordbruksverket.se/

webdav/files/SJV/trycksaker/Pdf_rapporter/ra12_20.pdf. Accessed in October 2012.

Bodin, Ö., and S. Saura. 2010. Ranking individual habitat patches as connectivity providers:

integrating network analysis and patch removal experiments. Ecological Modelling

221:2393-2405.

References

217

Boitani, L. 1995. Ecological and cultural diversities in the evolution of wolf-human relationships.

In: L. N. Carbyn, and D. R. Seip (eds). Ecology and conservation of wolves in a changing

world. Canadian Circumpolar Institute, Alberta, pp 3-11.

Bojarska, K., and N. Selva. 2012. Spatial patterns in brown bear Ursus arctos diet: the role of

geographical and environmental factors. Mammal Review 42:120-143.

Brambilla, M., F. Casale, V. Bergero, G. Bogliani, G. M. Crovetto, R. Falco, M. Roati, and I. Negri.

2010. Glorious past, uncertain present, bad future? Assessing effects of land-use changes

on habitat suitability for a threatened farmland bird species. Biological Conservation

143:2770-2778.

Brambilla, M., and P. Pedrini. 2013. The introduction of subsidies for grassland conservation in the

Italian Alps coincided with population decline in a threatened grassland species, the

Corncrake Crex crex. Bird Study 60:404-408.

Breitenmoser, U. 1998. Large predators in the Alps: the fall and rise of man's competitors.


Breitenmoser, U., C. Breitenmoser-Würsten, H. Okarma, T. Kaphegyi, U. Kaphygyi-Wallmann, U.

M. Müller. 2000. Action plan for the conservation of the Eurasian lynx in Europe (Lynx

lynx). Council of Europe, Strasbourg, 69 pp.

Brodier, S., S. Augiron, T. Cornulier, and V. Bretagnolle. 2013. Local improvement of skylark and

corn bunting population trends on intensive arable landscape: a case study of the

conservation tool Natura 2000. Animal Conservation 17:204-216.

Brook, L. A., C. N. Johnson, and E. G. Ritchie. 2012. Effects of predator control on behaviour of

an apex predator and indirect consequences for mesopredator suppression. Journal of

Applied Ecology 49:1278-1286.

Bucşa, C. and I. Tăuşan. 2010. Preliminary data on the sylophagous beetles (Insecta: Coleoptera)

from the "Breite ancient oak trees" nature reserve (Sighişoara, Romania). Brukenthal Acta

Musei 5:593-606.

Bucşa, L. 2007. Macromycetes of the Breite nature reserve of ancient oaks (Transylvania, Romania).

Transylvanian Review of Systematical and Ecological Research 4:33-42.

Budka, M., and T. S. Osiejuk. 2013. Habitat preferences of Corncrake (Crex crex) males in

agricultural meadows. Agriculture, Ecosystems & Environment 171:33-38.

Bugalho, M. N., M. C. Caldeira, J. S. Pereira, J. Aronson, and J. G. Pausas. 2011. Mediterranean

cork oak savannas require human use to sustain biodiversity and ecosystem services.

Frontiers in Ecology and the Environment 9:278-286.

Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference. Springer,

New York, 488 pp.

Butchart, S. H., M. Walpole, B. Collen, A. van Strien, J. P. Scharlemann, R. E. Almond, J. E. Baillie,

B. Bomhard, C. Brown, and J. Bruno. 2010. Global biodiversity: indicators of recent

declines. Science 328:1164-1168.

Buttler, A., F. Kohler, and F. Gillet. 2009. The Swiss mountain wooded pastures: patterns and

processes. In: A. Rigueiro-Rodríguez, J. McAdam, M. R. Mosquera-Losada (eds).

Agroforestry in Europe: current status and future prospects. Springer, Dordrecht, pp 377-

396.

Caledonian Partnership. 2003. Restoring Natura Forest Habitats. Highland Birchwoods, Munlochy,

56 pp.

Can, Ö. E., N. D'Cruze, D. L. Garshelis, J. Beecham, and D. W. Macdonald. 2014. Resolving

human‐bear conflict: a global survey of countries, experts and key factors. Conservation

Letters 7:501-513.

Carter, N. H., S. J. Riley, and J. Liu. 2012a. Utility of a psychological framework for carnivore

conservation. Oryx 46:525-535.

References

218

Carter, N. H., B. K. Shrestha, J. B. Karki, N. M. B. Pradhan, and J. Liu. 2012b. Coexistence

between wildlife and humans at fine spatial scales. Proceedings of the National Academy

of Sciences 109:15360-15365.

Chaideftou, E., C. A. Thanos, E. Bergmeier, A. S. Kallimanis, and P. Dimopoulos. 2011. The herb

layer restoration potential of the soil seed bank in an overgrazed oak forest. Journal of

Biological Research 15:47-57.

Chapron, G., P. Kaczensky, J. D. Linnell, M. von Arx, D. Huber, H. Andrén, J. V. López-Bao, M.

Adamec, F. Álvares, and O. Anders. 2014. Recovery of large carnivores in Europe’s

modern human-dominated landscapes. Science 346:1517-1519.

Chestin, I. 1999. The management of the brown bear in Romania. In: C. Servheen, S. Herrero, and

B. Peyton (eds). Bears: status survey and conservation action plan. International Union for

the Conservation of Nature and Natural Resources, Gland, pp 136-143.

Chételat, J., M. Kalbermatten, K. S. Lannas, T. Spiegelberger, J.-B. Wettstein, F. Gillet, A. Peringer,

and A. Buttler. 2013. A contextual analysis of land-use and vegetation changes in two

wooded pastures in the Swiss Jura Mountains. Ecology and Society 18:39.

Ciarniello, L. M., M. S. Boyce, D. C. Heard, and D. R. Seip. 2007. Components of grizzly bear

habitat selection: density, habitats, roads, and mortality risk. The Journal of Wildlife

Management 71:1446-1457.

Ciochia, V. 1992. Păsările clocitoare din România. Editura Ştiintifică, Bucharest, 386 pp.

Ciucci, P., and L. Boitani. 2008. The Apennine brown bear: a critical review of its status and

conservation problems. Ursus 19:130-145.

Clevenger, A. P., F. J. Purroy, and M. R. Pelton. 1992. Food habits of brown bears (Ursus arctos) in

the Cantabrian Mountains, Spain. Journal of Mammalogy 73:415-421.

Clevenger, A. P., and N. Waltho. 2001. Factors influencing the effectiveness of wildlife underpasses

in Banff National Park, Alberta, Canada. Conservation Biology 14:47-56.

Coldea, G. 1992. The vegetation of Transylvania and Maramureş. In: D. Ivan (ed). Vegetaţia

României. Editura Technică Agricolă, Bucharest, pp. 162-183 (in Romanian).

Colman, N., C. Gordon, M. Crowther, and M. Letnic. 2014. Lethal control of an apex predator has

unintended cascading effects on forest mammal assemblages. Proceedings of the Royal

Society B: Biological Sciences 281:20133094.

Concepción, E. D., M. Díaz, and R. A. Baquero. 2008. Effects of landscape complexity on the

ecological effectiveness of agri-environment schemes. Landscape Ecology 23:135-148.

Conner, R. N., and D. C. Rudolph. 1991. Forest habitat loss, fragmentation, and red-cockaded

woodpecker populations. Wilson Bulletin 103:446-457.

Corbett, P. E., and M. D. Hudson. 2010. Management of cover areas may increase numbers of

breeding Corncrakes Crex crex. Bird Study 57:553-559.

Costa, A., M. Madeira, J. Lima Santos, and Â. Oliveira. 2011. Change and dynamics in

Mediterranean evergreen oak woodlands landscapes of Southwestern Iberian Peninsula.

Landscape and Urban Planning 102:164-176.

Cramp, S. 2000. The complete birds of the western palearctic. CD-ROM edition. Oxford

University Press. Oxford.

Creel, S., and D. Christianson. 2008. Relationships between direct predation and risk effects.

Trends in Ecology & Evolution 23:194-201.

Cremene, C., G. Groza, L. Rakosy, A. A. Schileyko, A. Baur, A. Erhardt, and B. Baur. 2005.

Alterations of steppe-like grasslands in Eastern Europe: a threat to regional biodiversity

hotspots. Conservation Biology 19:1606-1618.

Crooks, K. R. 2002. Relative sensitivities of mammalian carnivores to habitat fragmentation.

Conservation Biology 16:488-502.

References

219

Crooks, K. R., C. L. Burdett, D. M. Theobald, C. Rondinini, and L. Boitani. 2011. Global patterns

of fragmentation and connectivity of mammalian carnivore habitat. Philosophical

Transactions of the Royal Society B: Biological Sciences 366:2642-2651.

Crooks, K. R., and M. E. Soulé. 1999. Mesopredator release and avifaunal extinctions in a

fragmented system. Nature 400:563-566.

Crutzen, P. J. 2002. Geology of mankind. Nature 415:23-23.

Dahle, B., O. J. Sørensen, E. H. Wedul, J. E. Swenson, and F. Sandegren. 1998. The diet of brown

bears Ursus arctos in central Scandinavia: effect of access to free-ranging domestic sheep

Ovis aries. Wildlife Biology 4:147-158.

Dahlström, A., A. Iuga, and T. Lennartsson. 2013. Managing biodiversity rich hay meadows in the

EU: a comparison of Swedish and Romanian grasslands. Environmental Conservation

40:194-205.

Dallimer, M., K. J. Gaston, A. M. J. Skinner, N. Hanley, S. Acs, and P. R. Armsworth. 2010. Field-

level bird abundances are enhanced by landscape-scale agri-environment scheme uptake.

Biology Letters 6:643-646.

Dauber, J., J. Bengtsson, and L. Lenoir. 2006. Evaluating effects of habitat loss and land‐use

continuity on ant species richness in seminatural grassland remnants. Conservation Biology

20:1150-1160.

DeFries, R. S., J. A. Foley, and G. P. Asner. 2004. Land-use choices: balancing human needs and

ecosystem function. Frontiers in Ecology and the Environment 2:249-257.

Demeter, L., and A. Kelemen. 2012. Quantifying the abandonment of mountain hay meadows in

the Eastern Carpathians. European Forum on Nature Conservation and Pastoralism,

Hamsterley Vale, 21 pp.

Devictor, V., and F. Jiguet. 2007. Community richness and stability in agricultural landscapes: the

importance of surrounding habitats. Agriculture, Ecosystems & Environment 120:179-184.

Diacon-Bolli, J., T. Dalang, R. Holderegger, and M. Bürgi. 2012. Heterogeneity fosters biodiversity:

linking history and ecology of dry calcareous grasslands. Basic and Applied Ecology

13:641-653.

Diamond, J. 1986. Overview: laboratory experiments, field experiments, and natural experiments.

In: J. M. Diamond, and T. J. Case (eds). Community ecology. Harper & Row, New York,

pp 3-22.

Dickman, A. J. 2010. Complexities of conflict: the importance of considering social factors for

effectively resolving human-wildlife conflict. Animal Conservation 13:458-466.

Dickman, A. J., L. Hazzah, C. Carbone, and S. M. Durant. 2014. Carnivores, culture and

‘contagious conflict’: Multiple factors influence perceived problems with carnivores in

Tanzania’s Ruaha landscape. Biological Conservation 178:19-27.

Dickman, A. J., E. A. Macdonald, and D. W. Macdonald. 2011. A review of financial instruments

to pay for predator conservation and encourage human-carnivore coexistence. Proceedings

of the National Academy of Sciences 108:13937-13944.

Donadio, E., and S. W. Buskirk. 2006. Diet, morphology, and interspecific killing in Carnivora.

American Naturalist 167:524-536.

Donald, P., R. Green, and M. Heath. 2001. Agricultural intensification and the collapse of Europe's

farmland bird populations. Proceedings of the Royal Society of London Series B:

Biological Sciences 268:25-29.

Donald, P. F., and A. D. Evans. 2006. Habitat connectivity and matrix restoration: the wider

implications of agri-environment schemes. Journal of Applied Ecology 43:209-218.

Dormann, C. F., J. M. McPherson, M. B. Araújo, R. Bivand, J. Bolliger, G. Carl, R. G. Davies, A.

Hirzel, W. Jetz, W. Daniel Kissling, I. Kühn, R. Ohlemüller, P. R. Peres-Neto, B.

Reineking, B. Schröder, F. M. Schurr, and R. Wilson. 2007. Methods to account for spatial

References

220

autocorrelation in the analysis of species distributional data: a review. Ecography 30:609-

628.

Dorner, B. 1910. The agriculture of the Transylvanian Saxons. Győr, 336 pp (in Hungarian).

Dorresteijn, I., J. Hanspach, A. Kecskés, H. Latková, Z. Mezey, S. Sugár, H. von Wehrden, and J.

Fischer. 2014. Human-carnivore coexistence in a traditional rural landscape. Landscape

Ecology 29:1145-1155.

Dorresteijn, I., T. Hartel, J. Hanspach, H. von Wehrden, and J. Fischer. 2013. The conservation

value of traditional rural landscapes: the case of woodpeckers in Transylvania, Romania.

PLoS ONE 8:e65236.

Doxa, A., Y. Bas, M. L. Paracchini, P. Pointereau, J.-M. Terres, and F. Jiguet. 2010. Low-intensity

agriculture increases farmland bird abundances in France. Journal of Applied Ecology

47:1348-1356.

Dunning, J. B., B. J. Danielson, and H. R. Pulliam. 1992. Ecological processes that affect

populations in complex landscapes. Oikos:169-175.

Ellis, E. C., and N. Ramankutty. 2008. Putting people in the map: anthropogenic biomes of the

world. Frontiers in Ecology and the Environment 6:439-447.

Elmhagen, B., G. Ludwig, S. Rushton, P. Helle, and H. Lindén. 2010. Top predators,

mesopredators and their prey: interference ecosystems along bioclimatic productivity

gradients. Journal of Animal Ecology 79:785-794.

Elmhagen, B., and S. P. Rushton. 2007. Trophic control of mesopredators in terrestrial ecosystems:

top‐down or bottom‐up? Ecology Letters 10:197-206.

Elston, D., R. Moss, T. Boulinier, C. Arrowsmith, and X. Lambin. 2001. Analysis of aggregation, a

worked example: numbers of ticks on red grouse chicks. Parasitology 122:563-569.

Enserink, M., and G. Vogel. 2006. Wildlife conservation: the carnivore comeback. Science 314:746-

749.

Ernoult, A., and D. Alard. 2011. Species richness of hedgerow habitats in changing agricultural

landscapes: are α and γ diversity shaped by the same factors? Landscape Ecology 26:683-

696.

Estes, J. A., J. Terborgh, J. S. Brashares, M. E. Power, J. Berger, W. J. Bond, S. R. Carpenter, T. E.

Essington, R. D. Holt, J. B. C. Jackson, R. J. Marquis, L. Oksanen, T. Oksanen, R. T.

Paine, E. K. Pikitch, W. J. Ripple, S. A. Sandin, M. Scheffer, T. W. Schoener, J. B. Shurin,

A. R. E. Sinclair, M. E. Soule, R. Virtanen, and D. A. Wardle. 2011. Trophic downgrading

of planet Earth. Science 333:301-306.

European Environment Agency [EEA]. 2006. Corine land use cover 2006 seamless vector data

base. Copenhagen. www.eea.europa.eu/data-and-maps/data/clc-2006-vector-data-version-

2. Accessed 12 March 2014.

Fahrig, L. 2003. Effects of habitat fragmentation on biodiversity. Annual Review of Ecology

Evolution and Systematics 34:487-515.

Farm Environment Plan Guide. 2006. Identifying ancient trees. Rural Development Service,

London, 2 pp.

Fernández, N., N. Selva, C. Yuste, H. Okarma, and Z. Jakubiec. 2012. Brown bears at the edge:

Modeling habitat constrains at the periphery of the Carpathian population. Biological


Ferraro, P. J., and A. Kiss. 2002. Direct payments to conserve biodiversity. Science 298:1718-1719.

Fielding, A. H., and J. F. Bell. 1997. A review of methods for the assessment of prediction errors in

conservation presence/absence models. Environmental Conservation 24:38-49.

Filippi-Codaccioni, O., V. Devictor, Y. Bas, J. Clobert, and R. Julliard. 2010. Specialist response to

proportion of arable land and pesticide input in agricultural landscapes. Biological


References

221

Fischer, C., A. Flohre, L. W. Clement, P. Batary, W. W. Weisser, T. Tscharntke, and C. Thies. 2011.

Mixed effects of landscape structure and farming practice on bird diversity. Agriculture

Ecosystems & Environment 141:119-125.

Fischer, J., B. Brosi, G. C. Daily, P. R. Ehrlich, R. Goldman, J. Goldstein, D. B. Lindenmayer, A.

D. Manning, H. A. Mooney, and L. Pejchar. 2008. Should agricultural policies encourage

land sparing or wildlife-friendly farming? Frontiers in Ecology and the Environment 6:380-

385.

Fischer, J., T. Hartel, and T. Kuemmerle. 2012. Conservation policy in traditional farming

landscapes. Conservation Letters 5:167-175.

Fischer, J., and D. B. Lindenmayer. 2007. Landscape modification and habitat fragmentation: a

synthesis. Global Ecology and Biogeography 16:265-280.

Fischer, J., K. Sherren, J. Stott, A. Zerger, G. Warren, and J. Stein. 2009. Toward landscape-wide

conservation outcomes in Australia's temperate grazing region. Frontiers in Ecology and

the Environment 8:69-74.

Fischer, J., J. Stott, and B. S. Law. 2010. The disproportionate value of scattered trees. Biological


Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, F. S. Chapin, M. T.

Coe, G. C. Daily, H. K. Gibbs, J. H. Helkowski, T. Holloway, E. A. Howard, C. J.

Kucharik, C. Monfreda, J. A. Patz, I. C. Prentice, N. Ramankutty, and P. K. Snyder. 2005.

Global consequences of land use. Science 309:570-574.

Folke, C. 2006. Resilience: the emergence of a perspective for social-ecological systems analyses.

Global Environmental Change 16:253-267.

Folke, C., S. Carpenter, B. Walker, M. Scheffer, T. Elmqvist, L. Gunderson, and C. Holling. 2004.

Regime shifts, resilience, and biodiversity in ecosystem management. Annual Review of

Ecology, Evolution, and Systematics 35:557-581.

Fourcade, Y., J. O. Engler, A. G. Besnard, D. Rödder, and J. Secondi. 2013. Confronting expert-

based and modelled distributions for species with uncertain conservation status: a case

study from the corncrake (Crex crex). Biological Conservation 167:161-171.

Gabriel, D., S. M. Sait, J. A. Hodgson, U. Schmutz, W. E. Kunin, and T. G. Benton. 2010. Scale

matters: the impact of organic farming on biodiversity at different spatial scales. Ecology

Letters 13:858-869.

Gangaas, K. E., B. P. Kaltenborn, and H. P. Andreassen. 2013. Geo-spatial aspects of acceptance

of illegal hunting of large carnivores in Scandinavia. PLoS ONE 8:e68849.

Garbarino, M., E. Lingua, M. M. Subira, and R. Motta. 2011. The larch wood pasture: structure and

dynamics of a cultural landscape. European Journal of Forest Research 130:491-502.

Gaston, K. J., T. M. Blackburn, and K. K. Goldewijk. 2003. Habitat conversion and global avian

biodiversity loss. Proceedings of the Royal Society B: Biological Sciences 270:1293-1300.

Gehring, T. M., K. C. VerCauteren, and J.-M. Landry. 2010. Livestock protection dogs in the 21st

century: is an ancient tool relevant to modern conservation challenges? BioScience 60:299-

308.

Geiger, F., J. Bengtsson, F. Berendse, W. W. Weisser, M. Emmerson, M. B. Morales, P. Ceryngier,

J. Liira, T. Tscharntke, and C. Winqvist. 2010. Persistent negative effects of pesticides on

biodiversity and biological control potential on European farmland. Basic and Applied

Ecology 11:97-105.

Gelman, A., Y. S. SU, V. Dorie, and M. Y. S. SU. 2013. Package ‘arm’. R package version 1.7-7.

Ghosal, S., V. Athreya, J. C. Linnell, and P. Vedeld. 2013. An ontological crisis? A review of large

felid conservation in India. Biodiversity and Conservation 22:2665-2681.

Gibbons, P., and D. Lindenmayer 2003. Tree hollows and wildlife conservation in Australia.

CSIRO, Collingwood, 240 pp.

References

222

Glikman, J. A., and B. Frank. 2011. Human dimensions of wildlife in Europe: the Italian way.

Human Dimensions of Wildlife 16:368-377.

Glikman, J. A., J. J. Vaske, A. J. Bath, P. Ciucci, and L. Boitani. 2012. Residents’ support for wolf

and bear conservation: the moderating influence of knowledge. European Journal of

Wildlife Research 58:295-302.

Godvik, I. M. R., L. E. Loe, J. O. Vik, V. Veiberg, R. Langvatn, and A. Mysterud. 2009. Temporal

scales, trade-offs, and functional responses in red deer habitat selection. Ecology 90:699-

710.

Goldberg, E., K. Kirby, J. Hall, and J. Latham. 2007. The ancient woodland concept as a practical

conservation tool in Great Britain. Journal for Nature Conservation 15:109-119.

Gonçalves, P., S. Alcobia, L. Simões, and M. Santos-Reis. 2012. Effects of management options on

mammal richness in a Mediterranean agro-silvo-pastoral system. Agroforestry Systems

85:383-395.

Goudie, A. S. 2013. The human impact on the natural environment: past, present, and future.

Wiley-Blackwell, Malden, 422 pp.

Grabovsky, V. 1993. Spatial distribution and spacing behaviour of males in a Russian Corncrake

(Crex crex) population. Gibier faune sauvage 10:259-279.

Grace, J. B. 2006. Structural equation modeling and natural systems. Cambridge University Press,

Cambridge, 378 pp.

Grace, J. B., D. R. Schoolmaster Jr, G. R. Guntenspergen, A. M. Little, B. R. Mitchell, K. M. Miller,

and E. W. Schweiger. 2012. Guidelines for a graph-theoretic implementation of structural

equation modeling. Ecosphere 3:art73.

Green, R. E., G. Rocamora, and N. Schaffer. 1997. Populations, ecology and threats to the

Corncrake Crex crex in Europe. Vogelwelt Berlin 118:117-134.

Gregory, R. D., A. Van Strien, P. Vorisek, A. W. G. Meyling, D. G. Noble, R. P. Foppen, and D.

W. Gibbons. 2005. Developing indicators for European birds. Philosophical Transactions

of the Royal Society B: Biological Sciences 360:269-288.

Griffiths, P., T. Kuemmerle, M. Baumann, V. C. Radeloff, I. V. Abrudan, J. Lieskovsky, C.

Munteanu, K. Ostapowicz, and P. Hostert. 2013. Forest disturbances, forest recovery, and

changes in forest types across the Carpathian ecoregion from 1985 to 2010 based on

Landsat image composites. Remote Sensing of Environment 151: 72-88.

Gruber, B., D. Evans, K. Henle, B. Bauch, D. Schmeller, F. Dziock, P.-Y. Henry, S. Lengyel, C.

Margules, and C. Dormann. 2012. ‘Mind the gap!’: how well does Natura 2000 cover

species of European interest? Nature Conservation 3:45-62.

Guerrero, I., M. B. Morales, J. J. Oñate, F. Geiger, F. Berendse, G. d. Snoo, S. Eggers, T. Pärt, J.

Bengtsson, L. W. Clement, W. W. Weisser, A. Olszewski, P. Ceryngier, V. Hawro, J. Liira,

T. Aavik, C. Fischer, A. Flohre, C. Thies, and T. Tscharntke. 2012. Response of ground-

nesting farmland birds to agricultural intensification across Europe: landscape and field

level management factors. Biological Conservation 152:74-80.

Gusset, M., A. H. Maddock, G. J. Gunther, M. Szykman, R. Slotow, M. Walters, and M. J. Somers.

2008. Conflicting human interests over the re-introduction of endangered wild dogs in

South Africa. Biodiversity and Conservation 17:83-101.

Güthlin, D., F. Knauer, T. Kneib, H. Küchenhoff, P. Kaczensky, G. Rauer, M. Jonozovič, A.

Mustoni, and K. Jerina. 2011. Estimating habitat suitability and potential population size

for brown bears in the Eastern Alps. Biological Conservation 144:1733-1741.

Haberl, H., K. H. Erb, F. Krausmann, V. Gaube, A. Bondeau, C. Plutzar, S. Gingrich, W. Lucht,

and M. Fischer-Kowalski. 2007. Quantifying and mapping the human appropriation of net

primary production in Earth's terrestrial ecosystems. Proceedings of the National Academy

of Sciences 104:12942-12947.

References

223

Halada, L., D. Evans, C. Romão, and J. E. Petersen. 2011. Which habitats of European importance

depend on agricultural practices? Biodiversity and Conservation 20:2365-2378.

Hamel, S., S. T. Killengreen, J. A. Henden, N. E. Eide, L. Roed‐Eriksen, R. A. Ims, and N. G.

Yoccoz. 2012. Towards good practice guidance in using camera‐traps in ecology: influence

of sampling design on validity of ecological inferences. Methods in Ecology and Evolution

4:105-113.

Hanski, I. 2011. Habitat loss, the dynamics of biodiversity, and a perspective on conservation.

Ambio 40:248-255.

Hanspach, J., J. Fischer, J. Stott, and K. Stagoll. 2011. Conservation management of eastern

Australian farmland birds in relation to landscape gradients. Journal of Applied Ecology

48:523-531.

Hanspach, J., T. Hartel, A. I. Milcu, F. Mikulcak, I. Dorresteijn, J. Loos, H. von Wehrden, T.

Kuemmerle, D. Abson, A. Kovács-Hostyánszki, A. Báldi, and J. Fischer. 2014. A holistic

approach to studying social-ecological systems and its application to southern

Transylvania. Ecology and Society 19:32.

Haraszthy, L., F. Márkus, and L. Bank. 1997. The nature conservation of wood-pastures. WWF-

füzetek 12, 23 pp (in Hungarian).

Hartel, T., R. Bancila, and D. Cogalniceanu. 2011. Spatial and temporal variability of aquatic habitat

use by amphibians in a hydrologically modified landscape. Freshwater Biology 56:2288-

2298.

Hartel, T., I. Dorresteijn, C. Klein, O. Máthé, C. I. Moga, K. Öllerer, M. Roellig, H. von Wehrden,

and J. Fischer. 2013. Wood-pastures in a traditional rural region of Eastern Europe:

characteristics, management and status. Biological Conservation 166:267-275.

Hartel, T., J. Fischer, C. Câmpeanu, A. I. Milcu, J. Hanspach, and I. Fazey. 2014. The importance

of ecosystem services for rural inhabitants in a changing cultural landscape in Romania.

Ecology and Society 19:42.

Hartel, T., and C. I. Moga. 2010. Good practice management of wood-pasture habitats. Fundatia

Mihai Eminescu Trust, Sighisoara (in Romanian).

Hazzah, L., M. Borgerhoff Mulder, and L. Frank. 2009. Lions and warriors: social factors

underlying declining African lion populations and the effect of incentive-based

management in Kenya. Biological Conservation 142:2428-2437.

Heinlein, K., R. Schöller, M. Scholz, and R. Wölfel 2005. Hutanger: Natur- und Kulturerbe mit

Zukunft. Pfeifer, 224 pp (in German).

Hemson, G., S. Maclennan, G. Mills, P. Johnson, and D. Macdonald. 2009. Community, lions,

livestock and money: a spatial and social analysis of attitudes to wildlife and the

conservation value of tourism in a human-carnivore conflict in Botswana. Biological


Henle, K., D. Alard, J. Clitherow, P. Cobb, L. Firbank, T. Kull, D. McCracken, R. F. A. Moritz, J.

Niemelae, M. Rebane, D. Wascher, A. Watt, and J. Young. 2008. Identifying and managing

the conflicts between agriculture and biodiversity conservation in Europe: A review.

Agriculture Ecosystems & Environment 124:60-71.

Herzon, I., A. Auninš, J. Elts, and Z. Preikša. 2008. Intensity of agricultural land-use and farmland

birds in the Baltic States. Agriculture, Ecosystems & Environment 125:93-100.

Hinsley, S., and P. Bellamy. 2000. The influence of hedge structure, management and landscape

context on the value of hedgerows to birds: a review. Journal of Environmental


Hinsley, S. A. 2000. The costs of multiple patch use by birds. Landscape Ecology 15:765-775.

References

224

Höchtl, F., S. Lehringer, and W. Konold. 2005. “Wilderness”: what it means when it becomes a

reality—a case study from the southwestern Alps. Landscape and Urban Planning 70:85-

95.

Hole, D., A. Perkins, J. Wilson, I. Alexander, P. Grice, and A. Evans. 2005. Does organic farming

benefit biodiversity? Biological Conservation 122:113-130.

Holmern, T., J. Nyahongo, and E. Røskaft. 2007. Livestock loss caused by predators outside the

Serengeti National Park, Tanzania. Biological Conservation 135:518-526.

Holzwarth, F., A. Kahl, J. Bauhus, and C. Wirth. 2013. Many ways to die: partitioning tree mortality

dynamics in a near‐natural mixed deciduous forest. Journal of Ecology 101:220-230.

Horák, J., and K. Rébl. 2013. The species richness of click beetles in ancient pasture woodland

benefits from a high level of sun exposure. Journal of Insect Conservation 17:307-318.

Huber, D. 2007. Ursus arctos. The IUCN Red List of Threathened Species, version 2013.2.

International Union for Conservation of Nature and Natural Resources.

www.iucnredlist.org. Accessed 12 Mar 2014.

Huber, D., and H. U. Roth. 1993. Movements of European brown bears in Croatia. Acta

Theriologica 38:151-159.

Hughes, J., and D. W. Macdonald. 2013. A review of the interactions between free-roaming

domestic dogs and wildlife. Biological Conservation 157:341-351.

Huntington, H. P. 2000. Using traditional ecological knowledge in science: methods and

applications. Ecological Applications 10:1270-1274.

Inger, R., R. Gregory, J. P. Duffy, I. Stott, P. Voříšek, and K. J. Gaston. 2015. Common European

birds are declining rapidly while less abundant species' numbers are rising. Ecology Letters

18:28-36.

INS [Institutul National de Statistica; Romanian National Institute of Statistics]. 2011. The 2011

population and housing census (database at the level of administrative units). Bucharest.

Received 6 February 2012.

Inskip, C., Z. Fahad, R. Tully, T. Roberts, and D. MacMillan. 2014. Understanding carnivore killing

behaviour: exploring the motivations for tiger killing in the Sundarbans, Bangladesh.


Ives, C. D., and D. Kendal. 2014. The role of social values in the management of ecological

systems. Journal of Environmental Management 144:67-72.

Jȩdrzejewski, W., B. Jȩdrzejewska, H. Okarma, K. Schmidt, K. Zub, and M. Musiani. 2000. Prey

Selection and predation by wolves in Bialowieza Primeval Forest, Poland. Journal of

Mammalogy 81:197-212.

Jedrzejewski, W., M. Niedzialkowska, S. Nowak, and B. Jedrzejewska. 2004. Habitat variables

associated with wolf (Canis lupus) distribution and abundance in northern Poland. Diversity

and Distributions 10:225-233.

Jedrzejewski, W., K. Schmidt, J. Theuerkauf, B. Jedrzejewska, N. Selva, K. Zub, and L. Szymura.

2002. Kill rates and predation by wolves on ungulate populations in Bialowieza Primeval

Forest (Poland). Ecology 83:1341-1356.

Johnson, C. N., and J. VanDerWal. 2009. Evidence that dingoes limit abundance of a

mesopredator in eastern Australian forests. Journal of Applied Ecology 46:641-646.

Josefsson, J., Å. Berg, M. Hiron, T. Pärt, and S. Eggers. 2013. Grass buffer strips benefit

invertebrate and breeding skylark numbers in a heterogeneous agricultural landscape.


Kaczensky, P. 1999. Large carnivore depredation on livestock in Europe. Ursus 11:59-71.

Kaczensky, P., M. Blazic, and H. Gossow. 2004. Public attitudes towards brown bears (Ursus arctos)

in Slovenia. Biological Conservation 118:661-674.

References

225

Kaczensky, P., K. Jerina, M. Jonozovič, M. Krofel, T. Skrbinšek, G. Rauer, I. Kos, and B. Gutleb.

2011. Illegal killings may hamper brown bear recovery in the Eastern Alps. Ursus 22:37-46.

Kaczensky, P., F. Knauer, B. Krze, M. Jonozovic, M. Adamic, and H. Gossow. 2003. The impact

of high speed, high volume traffic axes on brown bears in Slovenia. Biological


Kansky, R., M. Kidd, and A. T. Knight. 2014. Meta‐analysis of attitudes toward damage‐causing

mammalian wildlife. Conservation Biology 28:924-938.

Kaplan, J. O., K. M. Krumhardt, and N. Zimmermann. 2009. The prehistoric and preindustrial

deforestation of Europe. Quaternary Science Reviews 28:3016-3034.

Karamanlidis, A. A., A. Sanopoulos, L. Georgiadis, and A. Zedrosser. 2011. Structural and

economic aspects of human-bear conflicts in Greece. Ursus 22:141-151.

Karamanlidis, A. A., M. Straka, E. Drosopoulou, M. de Gabriel Hernando, I. Kocijan, L. Paule, and

Z. Scouras. 2012. Genetic diversity, structure, and size of an endangered brown bear

population threatened by highway construction in the Pindos Mountains, Greece.

European Journal of Wildlife Research 58:511-522.

Karanth, K. K., J. D. Nichols, K. U. Karanth, J. E. Hines, and N. L. Christensen. 2010. The

shrinking ark: patterns of large mammal extinctions in India. Proceedings of the Royal

Society B: Biological Sciences 277:1971-1979.

Keišs, O. 2005. Impact of changes in agricultural land use on the Corncrake Crex crex population in

Latvia. Acta Univ Latv 691:93-109.

Kellert, S. R., M. Black, C. R. Rush, and A. J. Bath. 1996. Human culture and large carnivore

conservation in North America. Conservation Biology 10:977-990.

Kleijn, D., R. A. Baquero, Y. Clough, M. Díaz, J. De Esteban, F. Fernández, D. Gabriel, F. Herzog,

A. Holzschuh, R. Jöhl, E. Knop, A. Kruess, E. J. P. Marshall, I. Steffan-Dewenter, T.

Tscharntke, J. Verhulst, T. M. West, and J. L. Yela. 2006. Mixed biodiversity benefits of

agri-environment schemes in five European countries. Ecology Letters 9:243-254.

Kleijn, D., F. Kohler, A. Baldi, P. Batary, E. D. Concepcion, Y. Clough, M. Diaz, D. Gabriel, A.

Holzschuh, E. Knop, A. Kovacs, E. J. P. Marshall, T. Tscharntke, and J. Verhulst. 2009.

On the relationship between farmland biodiversity and land-use intensity in Europe.

Proceedings of the Royal Society B: Biological Sciences 276:903-909.

Kleijn, D., M. Rundlof, J. Scheper, H. G. Smith, and T. Tscharntke. 2011. Does conservation on

farmland contribute to halting the biodiversity decline? Trends in Ecology & Evolution

26:474-481.

Kleijn, D., and W. J. Sutherland. 2003. How effective are European agri-environment schemes in

conserving and promoting biodiversity? Journal of Applied Ecology 40:947-969.

Knorn, J., T. Kuemmerle, V. C. Radeloff, A. Szabo, M. Mindrescu, W. S. Keeton, I. Abrudan, P.

Griffiths, V. Gancz, and P. Hostert. 2012. Forest restitution and protected area

effectiveness in post-socialist Romania. Biological Conservation 146:204-212.

Knorn, J., A. Rabe, V. C. Radeloff, T. Kuemmerle, J. Kozak, and P. Hostert. 2009. Land cover

mapping of large areas using chain classification of neighboring Landsat satellite images.

Remote Sensing of Environment 113:957-964.

Koffijberg, K., and N. Schäffer 2006. International single species action plan for the conservation

of the corncrake Crex crex. CMS Technical Series 14 & AEWA Technical Series 9, Bonn,

53 pp.

Koleček, J., M. Schleuning, I. J. Burfield, A. Báldi, K. Böhning-Gaese, V. Devictor, J. M.

Fernández-García, D. Hořák, C. A. Van Turnhout, and O. Hnatyna. 2014. Birds protected

by national legislation show improved population trends in Eastern Europe. Biological


References

226

Kopatz, A., H. G. Eiken, S. B. Hagen, M. Ruokonen, R. Esparza-Salas, J. Schregel, I. Kojola, M. E.

Smith, I. Wartiainen, and P. E. Aspholm. 2012. Connectivity and population subdivision at

the fringe of a large brown bear (Ursus arctos) population in North Western Europe.

Conservation Genetics 13:681-692.

Kosiński, Z., and M. Kempa. 2007. Density, distribution and nest-sites of woodpeckers picidae, in a

managed forest of Western Poland. Polish Journal of Ecology 55:519-533.

Krofel, M., M. Jonozovič, and K. Jerina. 2012. Demography and mortality patterns of removed

brown bears in a heavily exploited population. Ursus 23:91-103.

Kuemmerle, T., P. Hostert, V. Radeloff, S. Linden, K. Perzanowski, and I. Kruhlov. 2008. Cross-

border comparison of post-socialist farmland abandonment in the Carpathians.

Ecosystems 11:614-628.

Kuijper, D. 2011. Lack of natural control mechanisms increases wildlife-forestry conflict in

managed temperate European forest systems. European Journal of Forest Research

130:895-909.

Kuijper, D. P. J., C. de Kleine, M. Churski, P. van Hooft, J. Bubnicki, and B. Jędrzejewska. 2013.

Landscape of fear in Europe: wolves affect spatial patterns of ungulate browsing in

Białowieża Primeval Forest, Poland. Ecography 36:1263-1275.

Kuiper, M. W., H. J. Ottens, L. Cenin, A. P. Schaffers, J. van Ruijven, B. J. Koks, F. Berendse, and

G. R. de Snoo. 2013. Field margins as foraging habitat for skylarks (Alauda arvensis) in the

breeding season. Agriculture, Ecosystems & Environment 170:10-15.

Kumar, R., and P. Singh. 2010. Determining woodpecker diversity in the sub-Himalayan forests of

northern India using call playbacks. Journal of Field Ornithology 81:215-222.

Kurki, S., A. Nikula, P. Helle, and H. Lindén. 1998. Abundances of red fox and pine marten in

relation to the composition of boreal forest landscapes. Journal of Animal Ecology 67:874-

886.

Lagendijk, D. G., and M. Gusset. 2008. Human-carnivore coexistence on communal land

bordering the Greater Kruger Area, South Africa. Environmental management 42:971-976.

Laiolo, P., F. Dondero, E. Ciliento, and A. Rolando. 2004. Consequences of pastoral abandonment

for the structure and diversity of the alpine avifauna. Journal of Applied Ecology 41:294-

304.

Lammertink, M., D. M. Prawiradilaga, U. Setiorini, T. Z. Naing, J. W. Duckworth, and S. B. J.

Menken. 2009. Global population decline of the great slaty woodpecker (Mulleripicus

pulverulentus). Biological Conservation 142:166-179.

Laundré, J. W., L. Hernández, and K. B. Altendorf. 2001. Wolves, elk, and bison: reestablishing the

‘landscape of fear’ in Yellowstone National Park, USA. Canadian Journal of Zoology

79:1401-1409.

Lepers, E., E. F. Lambin, A. C. Janetos, R. DeFries, F. Achard, N. Ramankutty, and R. J. Scholes.

2005. A Synthesis of information on rapid land-cover change for the period 1981–2000.

BioScience 55:115-124.

Lescureux, N., J. Linnell, S. Mustafa, D. Melovski, A. Stojanov, G. Ivanov, and V. Avukatov. 2011.

The king of the forest: local knowledge about European brown bears (Ursus arctos) and

implications for their conservation in contemporary Western Macedonia. Conservation

and Society 9:189.

Lescureux, N., and J. D. Linnell. 2010. Knowledge and perceptions of Macedonian hunters and

herders: the influence of species specific ecology of bears, wolves, and lynx. Human

Ecology 38:389-399.

Lescureux, N., and J. D. Linnell. 2014. Warring brothers: the complex interactions between wolves

(Canis lupus) and dogs (Canis familiaris) in a conservation context. Biological Conservation

171:232-245.

References

227

Letnic, M., F. Koch, C. Gordon, M. S. Crowther, and C. R. Dickman. 2009. Keystone effects of an

alien top-predator stem extinctions of native mammals. Proceedings of the Royal Society

B: Biological Sciences 276:3249-3256.

Letnic, M., E. G. Ritchie, and C. R. Dickman. 2012. Top predators as biodiversity regulators: the

dingo Canis lupus dingo as a case study. Biological Reviews 87:390-413.

Liberg, O., G. Chapron, P. Wabakken, H. C. Pedersen, N. T. Hobbs, and H. Sand. 2012. Shoot,

shovel and shut up: cryptic poaching slows restoration of a large carnivore in Europe.


Lindborg, R., J. Bengtsson, Å. Berg, S. A. O. Cousins, O. Eriksson, T. Gustafsson, K. P. Hasund,

L. Lenoir, A. Pihlgren, E. Sjödin, and M. Stenseke. 2008. A landscape perspective on

conservation of semi-natural grasslands. Agriculture, Ecosystems & Environment 125:213-

222.

Lindenmayer, D. B., and J. Fischer 2006. Habitat fragmentation and landscape change: an

ecological and conservation synthesis. Island Press, Washington DC, 352 pp.

Lindenmayer, D. B., W. F. Laurance, J. F. Franklin, G. E. Likens, S. C. Banks, W. Blanchard, P.

Gibbons, K. Ikin, D. Blair, and L. McBurney. 2014. New policies for old trees: averting a

global crisis in a keystone ecological structure. Conservation Letters 7:61-69.

Linnell, J. D., C. Promberger, L. Boitani, J. E. Swenson, U. Breitenmoser, and R. Andersen. 2005.

The linkage between conservation strategies for large carnivores and biodiversity: the view

from the ‘half-full’ forests of Europe. In: J. C. Ray, K. H. Redford, R. Steneck, J. Berger

Large carnivores and the conservation of biodiversity. Island Press, Washington DC, pp

381-398.

Linnell, J. D. C., J. E. Swenson, and R. Anderson. 2001. Predators and people: conservation of

large carnivores is possible at high human densities if management policy is favourable.

Animal Conservation 4:345-349.

Linţia, D. 1954. Păsările din R.P.R., vol II. Ed. Acad. R.P.R., Bucharest, 298 pp.

Linţia, D. 1955. Păsările din R.P.R., vol III. Ed. Acad. R.P.R., Bucharest, 487 pp.

Liu, J. G., T. Dietz, S. R. Carpenter, M. Alberti, C. Folke, E. Moran, A. N. Pell, P. Deadman, T.

Kratz, J. Lubchenco, E. Ostrom, Z. Ouyang, W. Provencher, C. L. Redman, S. H.

Schneider, and W. W. Taylor. 2007. Complexity of coupled human and natural systems.

Science 317:1513-1516.

Liu, Y., M. Duan, and Z. Yu. 2013. Agricultural landscapes and biodiversity in China. Agriculture,


Llaneza, L., J. V. López‐Bao, and V. Sazatornil. 2012. Insights into wolf presence in human‐

dominated landscapes: the relative role of food availability, humans and landscape

attributes. Diversity and Distributions 18:459-469.

Long, R. A., T. M. Donovan, P. Mackay, W. J. Zielinski, and J. S. Buzas. 2007. Comparing scat

detection dogs, cameras, and hair snares for surveying carnivores. Journal of Wildlife


Loos, J., J. Hanspach, H. von Wehrden, M. Beldean, C. I. Moga, and J. Fischer. 2014. Developing

robust field survey protocols in landscape ecology: a case study on birds, plants and

butterflies. Biodiversity and Conservation, doi: 10.1007/s10531-014-0786-3.

Lovari, S., J. Herrero, J. Conroy, T. Maran, G. Giannatos, M. Stubbe, and S. Aulagnier. 2007. Cervus

elaphus. The IUCN Red List of Threatened Species. Version 2014.2. www.iucnredlist.org.

Accessed 04 November 2014.

Luick, R. 2008. Transhumance in Germany. Report to European Forum on Nature Conservation

and Pastoralism, Lampeter, 16 pp.

References

228

MacDonald, D., J. Crabtree, G. Wiesinger, T. Dax, N. Stamou, P. Fleury, J. Gutierrez Lazpita, and

A. Gibon. 2000. Agricultural abandonment in mountain areas of Europe: environmental

consequences and policy response. Journal of Environmental Management 59:47-69.

Maclennan, S. D., R. J. Groom, D. W. Macdonald, and L. G. Frank. 2009. Evaluation of a

compensation scheme to bring about pastoralist tolerance of lions. Biological Conservation

142:2419-2427.

Majić, A., and A. J. Bath. 2010. Changes in attitudes toward wolves in Croatia. Biological


Majić, A., A. Marino Taussig de Bodonia, Đ. Huber, and N. Bunnefeld. 2011. Dynamics of public

attitudes toward bears and the role of bear hunting in Croatia. Biological Conservation

144:3018-3027.

Manning, A. D., J. Fischer, and D. B. Lindenmayer. 2006. Scattered trees are keystone structures:

implications for conservation. Biological Conservation 132:311-321.

Manning, A. D., P. Gibbons, and D. B. Lindenmayer. 2009. Scattered trees: a complementary

strategy for facilitating adaptive responses to climate change in modified landscapes?

Journal of Applied Ecology 46:915-919.

Martin, J. M. J., M. B. M. Basille, B. Bram Van Moorter, V. Moorter, J. K. J. Kindberg, D. A. D.

Allainé, and J. E. S. J. Swenson. 2010. Coping with human disturbance: spatial and

temporal tactics of the brown bear (Ursus arctos). Canadian Journal of Zoology 88:875-883.

May, R., J. Van Dijk, P. Wabakken, J. E. Swenson, J. D. Linnell, B. Zimmermann, J. Odden, H. C.

Pedersen, R. Andersen, and A. Landa. 2008. Habitat differentiation within the large‐

carnivore community of Norway's multiple‐use landscapes. Journal of Applied Ecology

45:1382-1391.

Mazerolle, M. J. 2012. AICcmodavg: model selection and multimodel inference based on

(Q)AIC(c). R package version 1.24.

McGarigal, K., S. A. Cushman, and E. Ene. 2012. FRAGSTATS v4: spatial pattern analysis

program for categorical and continuous maps. Software, University of Massachusetts,

Amherst. www.umass.edu/landeco/research/fragstats/fragstats.html.

McGee, B. K., W. B. Ballard, K. L. Nicholson, B. L. Cypher, P. R. Lemons, and J. F. Kamler. 2006.

Effects of artificial escape dens on swift fox populations in northwest Texas. Wildlife

Society Bulletin 34:821-827.

McKenzie, A. J., S. B. Emery, J. R. Franks, and M. J. Whittingham. 2013. Landscape-scale

conservation: collaborative agri-environment schemes could benefit both biodiversity and

ecosystem services, but will farmers be willing to participate? Journal of Applied Ecology

50:1274-1280.

Mech, D. L. 2012. Is science in danger of sanctifying the wolf? Biological Conservation 150:143-

149.

Melis, C., B. Jędrzejewska, M. Apollonio, K. A. Bartoń, W. Jędrzejewski, J. D. Linnell, I. Kojola, J.

Kusak, M. Adamic, and S. Ciuti. 2009. Predation has a greater impact in less productive

environments: variation in roe deer, Capreolus capreolus, population density across Europe.

Global Ecology and Biogeography 18:724-734.

Mendenhall, C. D., D. S. Karp, C. F. Meyer, E. A. Hadly, and G. C. Daily. 2014. Predicting

biodiversity change and averting collapse in agricultural landscapes. Nature 509:213-217.

Merckx, T., R. E. Feber, P. Riordan, M. C. Townsend, N. A. D. Bourn, M. S. Parsons, and D. W.

Macdonald. 2009. Optimizing the biodiversity gain from agri-environment schemes.


Mertens, A., and C. Promberger. 2001. Economic aspects of large carnivore-livestock conflicts in

Romania. Ursus 12:173-180.

References

229

Mertens, A., and H. Schneider. 2005. What is wrong with Romanian livestock guarding dogs? A

discussion. Carnivore Damage Prevention News 9:9-14.

Mikulcak, F., J. L. Haider, D. J. Abson, J. Newig, and J. Fischer. 2015. Applying a capitals approach

to understand rural development traps: a case study from post-socialist Romania. Land Use

Policy 43:248-258.

Mikulcak, F., J. Newig, A. I. Milcu, T. Hartel, and J. Fischer. 2013. Integrating rural development

and biodiversity conservation in Central Romania. Environmental Conservation 40:1-9.

Mikusinski, G., and P. Angelstam. 1998. Economic geography, forest distribution, and woodpecker

diversity in central Europe. Conservation Biology 12:200-208.

Milcu, A. I., K. Sherren, J. Hanspach, D. Abson, and J. Fischer. 2014. Navigating conflicting

landscape aspirations: application of a photo-based Q-method in Transylvania (Central

Romania). Land Use Policy 41:408-422.

Millennium Ecosystem Assessment 2005. Ecosystems and human well-being: synthesis. Island

Press Washington, DC, 155 pp.

Ministry of Agriculture Forestry and Rural Development and the Ministry of Environment and

Water Management [MAPDR and MMGA]. 2005. Management and action plan for the

bear population in Romania. MAPDR and MMGA, Bucharest, 84 pp.

Mishra, C., P. Allen, T. McCarthy, M. Madhusudan, A. Bayarjargal, and H. H. Prins. 2003. The role

of incentive programs in conserving the snow leopard. Conservation Biology 17:1512-

1520.

Moga, C., T. Hartel, and K. Öllerer. 2009. Ancient oak wood-pasture as a habitat for the

endangered tree pipit Anthus trivialis. Biologia 64:1011-1015.

Moga, C. I., T. Hartel, and K. Öllerer. 2010. Status, microhabitat use and distribution of the

corncrake Crex crex in a Southern Transylvanian rural landscape, Romania. North-Western

Journal of Zoology 6:63-70.

Monastersky, R. 2014. Biodiversity: Life a status report. Nature 516:158-161.

Morell, V. 2013. Predators in the ‘hood. Science 341:1332-1335.

Morellet, N., B. Van Moorter, B. Cargnelutti, J.-M. Angibault, B. Lourtet, J. Merlet, S. Ladet, and A.

M. Hewison. 2011. Landscape composition influences roe deer habitat selection at both

home range and landscape scales. Landscape Ecology 26:999-1010.

Morris, A. J., and J. J. Gilroy. 2008. Close to the edge: predation risks for two declining farmland

passerines. Ibis 150:168-177.

Mountford, E. P., and G. E. Peterken. 2003. Long-term change and implications for the

management of woodpastures: experience over 40 years from Denny Wood, New Forest.

Forestry 76:19-43.

Muhly, T. B., M. Hebblewhite, D. Paton, J. A. Pitt, M. S. Boyce, and M. Musiani. 2013. Humans

strengthen bottom-up effects and weaken trophic cascades in a terrestrial food web. PLoS

ONE 8:e64311.

Munro, R. H. M., S. Nielsen, M. Price, G. Stenhouse, and M. Boyce. 2006. Seasonal and diel

patterns of grizzly bear diet and activity in west-central Alberta. Journal of Mammalogy

87:1112-1121.

Nakagawa, S., and H. Schielzeth. 2013. A general and simple method for obtaining R2 from

generalized linear mixed‐effects models. Methods in Ecology and Evolution 4:133-142.

Naughton-Treves, L., R. Grossberg, and A. Treves. 2003. Paying for tolerance: rural citizens'

attitudes toward wolf depredation and compensation. Conservation Biology 17:1500-1511.

Navarro, L. M., and H. M. Pereira. 2012. Rewilding abandoned landscapes in Europe. Ecosystems

15:900-912.

References

230

Naves, J., T. Wiegand, E. Revilla, and M. Delibes. 2003. Endangered species constrained by natural

and human factors: the case of brown bears in northern Spain. Conservation Biology

17:1276-1289.

Nellemann, C., O.-G. Støen, J. Kindberg, J. E. Swenson, I. Vistnes, G. Ericsson, J. Katajisto, B. P.

Kaltenborn, J. Martin, and A. Ordiz. 2007. Terrain use by an expanding brown bear

population in relation to age, recreational resorts and human settlements. Biological


Noyce, K. V., P. B. Kannowski, and M. R. Riggs. 1997. Black bears as ant-eaters: seasonal

associations between bear myrmecophagy and ant ecology in north-central Minnesota.

Canadian Journal of Zoology 75:1671-1686.

Oksanen, J., F. Guillaume, B. Kindt, R. Kindt, P. Legendre, P. R. Minchin, R. B. O'Hara, G. L.

Simpson, P. Solymos, M. H. H. Stevens, and H. Wagner. 2013. vegan: Community ecology

package. R package version 2.0-9.

Oksanen, L., and T. Oksanen. 2000. The logic and realism of the hypothesis of exploitation

ecosystems. American Naturalist 155:703-723.

Olea, L., and A. San Miguel-Ayanz. 2006. The Spanish dehesa: a traditional Mediterranean

silvopastoral system linking production and nature conservation. Grassland Science in

Europe 11:3-13.

Öllerer, K. 2012. The flora of the Breite wood-pasture (Sighisoara, Romania). Brukenthal Acta

Musei 7:589-604.

Öllerer, K. 2013. The vegetation of the Breite wood-pasture (Sighisoara, Romania): history, current

status and prospects. Brukenthal Acta Musei 8:547-566.

Olsen, W. 2004. Triangulation in social research: qualitative and quantitative methods can really be

mixed. Developments in Sociology 20:103-118.

Ordiz, A., R. Bischof, and J. E. Swenson. 2013. Saving large carnivores, but losing the apex

predator? Biological Conservation 168:128-133.

Oroszi, S. 2004. The forest management of Transylvanian Saxons/Erdészeti Egyesület.

Erdészettörténeti közlemények 63: 1-153 (in Hungarian).

Packer, C., A. Loveridge, S. Canney, T. Caro, S. T. Garnett, M. Pfeifer, K. K. Zander, A. Swanson,

D. MacNulty, G. Balme, H. Bauer, C. M. Begg, K. S. Begg, S. Bhalla, C. Bissett, T.

Bodasing, H. Brink, A. Burger, A. C. Burton, B. Clegg, S. Dell, A. Delsink, T. Dickerson,

S. M. Dloniak, D. Druce, L. Frank, P. Funston, N. Gichohi, R. Groom, C. Hanekom, B.

Heath, L. Hunter, H. H. DeIongh, C. J. Joubert, S. M. Kasiki, B. Kissui, W. Knocker, B.

Leathem, P. A. Lindsey, S. D. Maclennan, J. W. McNutt, S. M. Miller, S. Naylor, P. Nel, C.

Ng'weno, K. Nicholls, J. O. Ogutu, E. Okot-Omoya, B. D. Patterson, A. Plumptre, J.

Salerno, K. Skinner, R. Slotow, E. A. Sogbohossou, K. J. Stratford, C. Winterbach, H.

Winterbach, and S. Polasky. 2013. Conserving large carnivores: dollars and fence. Ecology

Letters 16:635-641.

Palang, H., A. Printsmann, E. K. Gyuro, M. Urbanc, E. Skowronek, and W. Woloszyn. 2006. The

forgotten rural landscapes of Central and Eastern Europe. Landscape Ecology 21:347-357.

Palomares, F., and T. M. Caro. 1999. Interspecific killing among mammalian carnivores. The

American Naturalist 153:492-508.

Paltto, H., A. Nordberg, B. Nordén, and T. Snäll. 2011. Development of secondary woodland in

oak wood pastures reduces the richness of rare epiphytic lichens. PLoS ONE 6:e24675.

Panek, M., and W. Bresiński. 2002. Red fox Vulpes vulpes density and habitat use in a rural area of

western Poland in the end of 1990s, compared with the turn of 1970s. Acta Theriologica

47:433-442.

Pasanen-Mortensen, M., M. Pyykönen, and B. Elmhagen. 2013. Where lynx prevail, foxes will fail:

limitation of a mesopredator in Eurasia. Global Ecology and Biogeography 22:868-877.

References

231

Pasinelli, G. 2000. Oaks (Quercus sp.) and only oaks? Relations between habitat structure and home

range size of the middle spotted woodpecker (Dendrocopos medius). Biological Conservation

93:227-235.

Patrut, A. 2011. Age estimation of ancient oaks in the Breite reserve. Report to the Mihai Eminescu

Trust (in Romanian).

Payne, C. J., E. G. Ritchie, L. T. Kelly, and D. G. Nimmo. 2014. Does fire influence the landscape-

scale distribution of an invasive mesopredator? PLoS ONE 9:e107862.

Pe'er, G., L. Dicks, P. Visconti, R. Arlettaz, A. Báldi, T. Benton, S. Collins, M. Dieterich, R.

Gregory, and F. Hartig. 2014. EU agricultural reform fails on biodiversity. Science

344:1090-1092.

Peake, T., and P. McGregor. 2001. Corncrake Crex crex census estimates: a conservation application

of vocal. Animal Biodiversity and Conservation 24:81-90.

Pellissier, V., J. Touroult, R. Julliard, J. Siblet, and F. Jiguet. 2013. Assessing the Natura 2000

network with a common breeding birds survey. Animal Conservation 16:566-574.

Pereira, H. M., P. W. Leadley, V. Proença, R. Alkemade, J. P. Scharlemann, J. F. Fernandez-

Manjarrés, M. B. Araújo, P. Balvanera, R. Biggs, and W. W. Cheung. 2010. Scenarios for

global biodiversity in the 21st century. Science 330:1496-1501.

Perlut, N. G., A. M. Strong, T. M. Donovan, and N. J. Buckley. 2006. Grassland songbirds in a

dynamic management landscape: behavioral responses and management strategies.

Ecological Applications 16:2235-2247.

Peterson, R. O., J. A. Vucetich, R. E. Page, and A. Chouinard. 2003. Temporal and spatial aspects

of predator-prey dynamics. Alces 39:215-232.

Pettersson, B. 1985. Relative importance of habitat area, isolation and quality for the occurrence of

middle spotted woodpecker Dendrocopos medius (L.) in Sweden. Ecography 8:53-58.

Pickett, S. R., and G. M. Siriwardena. 2011. The relationship between multi‐scale habitat

heterogeneity and farmland bird abundance. Ecography 34:955-969.

Piha, M., J. Tiainen, J. Holopainen, and V. Vepsäläinen. 2007. Effects of land-use and landscape

characteristics on avian diversity and abundance in a boreal agricultural landscape with

organic and conventional farms. Biological Conservation 140:50-61.

Pimentel, D., U. Stachow, D. A. Takacs, H. W. Brubaker, A. R. Dumas, J. J. Meaney, D. E. Onsi,

and D. B. Corzilius. 1992. Conserving biological diversity in agricultural/forestry systems.

BioScience 42:354-362.

Pimm, S. L., G. J. Russell, J. L. Gittleman, and T. M. Brooks. 1995. The future of biodiversity.

Science 269:347-349.

Plieninger, T. 2007. Compatibility of livestock grazing with stand regeneration in Mediterranean

holm oak parklands. Journal for Nature Conservation 15:1-9.

Plieninger, T., and C. Bieling. 2013. Resilience-based perspectives to guiding high nature value

farmland through socio-economic change. Ecology and Society 18:20.

Plieninger, T., F. Höchtl, and T. Spek. 2006. Traditional land-use and nature conservation in

European rural landscapes. Environmental Science & Policy 9:317-321.

Plieninger, T., and M. Schaar. 2008. Modification of land cover in a traditional agroforestry system

in Spain: processes of tree expansion and regression. Ecology and Society 13:25.

Polis, G. A., and R. D. Holt. 1992. Intraguild predation: the dynamics of complex trophic

interactions. Trends in Ecology & Evolution 7:151-154.

Popescu, V. D., L. Rozylowicz, I. M. Niculae, A. L. Cucu, and T. Hartel. 2014. Species, habitats,

society: an evaluation of research supporting EU's Natura 2000 Network. PLoS ONE

9:e113648.

References

232

Posillico, M., A. Meriggi, E. Pagnin, S. Lovari, and L. Russo. 2004. A habitat model for brown bear

conservation and land use planning in the central Apennines. Biological Conservation

118:141-150.

Pratt, M. G. 2009. For the lack of a boilerplate: tips on writing up (and reviewing) qualitative

research. Academy of Management Journal 52:856-862.

Preatoni, D., A. Mustoni, A. Martinoli, E. Carlini, B. Chiarenzi, S. Chiozzini, S. Van Dongen, L. A.

Wauters, and G. Tosi. 2005. Conservation of brown bear in the Alps: space use and

settlement behavior of reintroduced bears. Acta Oecologica 28:189-197.

Pretty, J. 2011. Interdisciplinary progress in approaches to address social-ecological and ecocultural

systems. Environmental Conservation 38:127-139.

Princé, K., J.-P. Moussus, and F. Jiguet. 2012. Mixed effectiveness of French agri-environment

schemes for nationwide farmland bird conservation. Agriculture, Ecosystems &

Environment 149:74-79.

Queiroz, C., R. Beilin, C. Folke, and R. Lindborg. 2014. Farmland abandonment: threat or

opportunity for biodiversity conservation? A global review. Frontiers in Ecology and the


Quelch, P. R. 2002. An illustrated guide to ancient wood-pasture in Scotland. Forestry

Commission, Lochgilphead, 54 pp.

Quinn, J. E., J. R. Brandle, and R. J. Johnson. 2012. The effects of land sparing and wildlife-friendly

practices on grassland bird abundance within organic farmlands. Agriculture, Ecosystems

& Environment 161:10-16.

R Development Core Team. 2013. R: a language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria.

Rackham, O. 1980. Ancient woodland, its history, vegetation and uses in England. Edward Arnold,

London, 402 pp.

Rackham, O. 1998. Savanna in Europe. In: K. J. Kirby, and C. Watkins (eds). The ecological history

of European forests, CAB International, Wallingford, pp 1-24.

Ranganathan, J., R. J. R. Daniels, M. D. S. Chandran, P. R. Ehrlich, and G. C. Daily. 2008.

Sustaining biodiversity in ancient tropical countryside. Proceedings of the National

Academy of Sciences 105:17852-17854.

Rathore, C. S., Y. Dubey, A. Shrivastava, P. Pathak, and V. Patil. 2012. Opportunities of habitat

connectivity for tiger (Panthera tigris) between Kanha and Pench National Parks in Madhya

Pradesh, India. PLoS ONE 7:e39996.

Ray, J. C., K. H. Redford, R. S. Steneck, and J. Berger. 2005. Large carnivores and the conservation

of biodiversity. Island Press, Washington DC, 526 pp.

Read, H. 2000. Veteran trees: a guide to good management. English Nature, Peterborough, 169 pp.

Redpath, S. M., J. Young, A. Evely, W. M. Adams, W. J. Sutherland, A. Whitehouse, A. Amar, R. A.

Lambert, J. D. Linnell, and A. Watt. 2012. Understanding and managing conservation

conflicts. Trends in Ecology & Evolution 28:100-109.

Reif, J., P. Voříšek, V. Bejcek, and J. Petr. 2008. Agricultural intensification and farmland birds:

new insights from a central European country. Ibis 150:596-605.

Renwick, A., T. Jansson, P. H. Verburg, C. Revoredo-Giha, W. Britz, A. Gocht, and D.

McCracken. 2013. Policy reform and agricultural land abandonment in the EU. Land Use

Policy 30:446-457.

Rigg, R. 2001. Livestock guarding dogs: their current use world wide. IUCN/SSC Canid Specialist

Group Occasional Paper No 1, 133 pp.

Rigg, R., S. Findo, M. Wechselberger, M. L. Gorman, C. Sillero-Zubiri, and D. W. Macdonald.

2011. Mitigating carnivore-livestock conflict in Europe: lessons from Slovakia. Oryx

45:272-280.

References

233

Ripple, W. J., and R. L. Beschta. 2012a. Large predators limit herbivore densities in northern forest

ecosystems. European Journal of Wildlife Research 58:733-742.

Ripple, W. J., and R. L. Beschta. 2012b. Trophic cascades in Yellowstone: the first 15 years after

wolf reintroduction. Biological Conservation 145:205-213.

Ripple, W. J., J. A. Estes, R. L. Beschta, C. C. Wilmers, E. G. Ritchie, M. Hebblewhite, J. Berger, B.

Elmhagen, M. Letnic, and M. P. Nelson. 2014. Status and ecological effects of the world’s

largest carnivores. Science 343:1241484.

Ripple, W. J., A. J. Wirsing, C. C. Wilmers, and M. Letnic. 2013. Widespread mesopredator effects

after wolf extirpation. Biological Conservation 160:70-79.

Ritchie, E. G., C. R. Dickman, M. Letnic, and A. T. Vanak. 2014. Dogs as predators and trophic

regulators. In: M. Gommper (ed). Free-ranging dogs and wildlife conservation. Oxford

University Press, Oxford, 55-68 pp.

Ritchie, E. G., B. Elmhagen, A. S. Glen, M. Letnic, G. Ludwig, and R. A. McDonald. 2012.

Ecosystem restoration with teeth: what role for predators? Trends in Ecology & Evolution

27:265-271.

Ritchie, E. G., and C. N. Johnson. 2009. Predator interactions, mesopredator release and

biodiversity conservation. Ecology Letters 12:982-998.

Roberge, J. M., and P. Angelstam. 2006. Indicator species among resident forest birds: a cross-

regional evaluation in northern Europe. Biological Conservation 130:134-147.

Roberge, J. M., P. Angelstam, and M. A. Villard. 2008. Specialised woodpeckers and naturalness in

hemiboreal forests: deriving quantitative targets for conservation planning. Biological


Robinson, S. K., F. R. Thompson, T. M. Donovan, D. R. Whitehead, and J. Faaborg. 1995.

Regional forest fragmentation and the nesting success of migratory birds. Science

267:1987-1990.

Robles, H., C. Ciudad, and E. Matthysen. 2011. Tree-cavity occurrence, cavity occupation and

reproductive performance of secondary cavity-nesting birds in oak forests: the role of

traditional management practices. Forest Ecology and Management 261:1428-1435.

Robles, H., C. Ciudad, R. Vera, P. P. Olea, F. J. Purroy, and E. Matthysen. 2007. Sylvopastoral

management and conservation of the middle spotted woodpecker at the south-western

edge of its distribution range. Forest Ecology and Management 242:343-352.

Robson, J. P., and F. Berkes. 2011. Exploring some of the myths of land use change: can rural to

urban migration drive declines in biodiversity? Global Environmental Change 21:844-854.

Rockström, J., W. Steffen, K. Noone, Å. Persson, F. S. Chapin, E. F. Lambin, T. M. Lenton, M.

Scheffer, C. Folke, and H. J. Schellnhuber. 2009. A safe operating space for humanity.

Nature 461:472-475.

Rolstad, J., B. Løken, and E. Rolstad. 2000. Habitat selection as a hierarchical spatial process: the

green woodpecker at the northern edge of its distribution range. Oecologia 124:116-129.

Rolstad, J., and E. Rolstad. 1995. Seasonal patterns in home range and habitat use of the grey-

headed woodpecker Picus canus as influenced by the availability of food. Ornis Fennica

72:1-13.

Rosenthal, G., J. Schrautzer, and C. Eichberg. 2012. Low-intensity grazing with domestic

herbivores: a tool for maintaining and restoring plant diversity in temperate Europe.

Teuxenia 32:167-205.

Roskaft, E., T. Bjerke, B. Kaltenborn, J. D. C. Linnell, and R. Andersen. 2003. Patterns of self-

reported fear towards large carnivores among the Norwegian public. Evolution and

Human Behavior 24:184-198.

References

234

Rudel, T. K., O. T. Coomes, E. Moran, F. Achard, A. Angelsen, J. Xu, and E. Lambin. 2005. Forest

transitions: towards a global understanding of land use change. Global Environmental

Change 15:23-31.

Sala, O. E., F. S. Chapin, J. J. Armesto, E. Berlow, J. Bloomfield, R. Dirzo, E. Huber-Sanwald, L. F.

Huenneke, R. B. Jackson, and A. Kinzig. 2000. Global biodiversity scenarios for the year

2100. Science 287:1770-1774.

Šálek, M., L. Drahníková, and E. Tkadlec. 2014. Changes in home range sizes and population

densities of carnivore species along the natural to urban habitat gradient. Mammal Review

45:1-14.

Salvatori, V., H. Okarma, O. Ionescu, Y. Dovhanych, S. Find'o, and L. Boitani. 2002. Hunting

legislation in the Carpathian Mountains: implications for the conservation and

management of large carnivores. Wildlife Biology 8:3-10.

Samelius, G., H. Andrén, P. Kjellander, and O. Liberg. 2013. Habitat selection and risk of

predation: re-colonization by lynx had limited impact on habitat selection by roe deer.

PLoS ONE 8:e75469.

Sanderson, E. W., M. Jaiteh, M. A. Levy, K. H. Redford, A. V. Wannebo, and G. Woolmer. 2002.

The human footprint and the last of the wild. BioScience 52:891-904.

Sanderson, F. J., A. Kloch, K. Sachanowicz, and P. F. Donald. 2009. Predicting the effects of

agricultural change on farmland bird populations in Poland. Agriculture, Ecosystems &


Sanderson, F. J., M. Kucharz, M. Jobda, and P. F. Donald. 2013. Impacts of agricultural

intensification and abandonment on farmland birds in Poland following EU accession.


Saura, S., and J. Torne. 2009. Conefor Sensinode 2.2: a software package for quantifying the

importance of habitat patches for landscape connectivity. Environmental Modelling &

Software 24:135-139.

Schmidt, K. T., and H. Hoi. 2002. Supplemental feeding reduces natural selection in juvenile red

deer. Ecography 25:265-272.

Schuette, P., S. Creel, and D. Christianson. 2013. Coexistence of African lions, livestock, and

people in a landscape with variable human land use and seasonal movements. Biological


Sergio, F., O. J. Schmitz, C. J. Krebs, R. D. Holt, M. R. Heithaus, A. J. Wirsing, W. J. Ripple, E.

Ritchie, D. Ainley, D. Oro, Y. Jhala, F. Hiraldo, and E. Korpimäki. 2014. Towards a

cohesive, holistic view of top predation: a definition, synthesis and perspective. Oikos

123:1234-1243.

Shirley, M. D., B. Elmhagen, P. W. Lurz, S. P. Rushton, and A. Angerbjörn. 2009. Modelling the

spatial population dynamics of arctic foxes: the effects of red foxes and microtine cycles.

Canadian Journal of Zoology 87:1170-1183.

Shurulinkov, P., G. Stoyanov, E. Komitov, G. Daskalova, and A. Ralev. 2012. Contribution to the

knowledge on distribution, number and habitat preferences of rare and endangered birds

in western Rhodopes Mountains, southern Bulgaria. Strigiformes and Piciformes. Acta

Zoologica Bulgarica 64:43-56.

Sirami, C., L. Brotons, I. Burfield, J. Fonderflick, and J.-L. Martin. 2008. Is land abandonment

having an impact on biodiversity? A meta-analytical approach to bird distribution changes

in the north-western Mediterranean. Biological Conservation 141:450-459.

Sklíba, J., and R. Fuchs. 2004. Male Corncrakes Crex crex extend their home ranges by visiting the

territories of neighbouring males: capsule radiotracked male Corncrake often intruded on

the territories of neighbouring males. Bird Study 51:113-118.

Skogen, K., I. Mauz, and O. Krange. 2008. Cry wolf!: narratives of wolf recovery in France and

Norway. Rural Sociology 73:105-133.

References

235

Skogstad G, A.Verdun. 2010. The Common Agricultural Policy: Policy Dynamics in a Changing

Context. Routledge, London, 176 pp.

Slobodyan, A. A. 1974. The European brown bear in the Carpathians. International Conference on

Bear Research and Management 3:313-319.

Smit, C., and J. L. Ruifrok. 2011. From protege to nurse plant: establishment of thorny shrubs in

grazed temperate woodlands. Journal of Vegetation Science 22:377-386.

Smit, C., and M. Verwijmeren. 2011. Tree-shrub associations in grazed woodlands: first rodents,

then cattle? Plant Ecology 212:483-493.

Smith, B. D. 2007. The ultimate ecosystem engineers. Science 315:1797-1798.

Söderström, B., T. Pärt, and E. Linnarsson. 2001. Grazing effects on between-year variation of

farmland bird communities. Ecological Applications 11:1141-1150.

Soulé, M. E., J. A. Estes, J. Berger, and C. M. Del Rio. 2003. Ecological effectiveness: conservation

goals for interactive species. Conservation Biology 17:1238-1250.

Spassov, N., and G. Spiridonov. 1999. The management of the brown bear in Romania. In: C.

Servheen, S. Herrero, and B. Peyton (eds). Bears: status survey and conservation action

plan. International Union for the Conservation of Nature and Natural Resources, Gland,

pp 59-63.

Spencer, J., and K. Kirby. 1992. An inventory of ancient woodland for England and Wales.


Steffen, W., P. J. Crutzen, and J. R. McNeill. 2007. The Anthropocene: are humans now

overwhelming the great forces of nature? Ambio 36:614-621.

Stein, A., K. Gerstner, and H. Kreft. 2014. Environmental heterogeneity as a universal driver of

species richness across taxa, biomes and spatial scales. Ecology Letters 17:866-880.

Stoate, C., A. Báldi, P. Beja, N. D. Boatman, I. Herzon, A. van Doorn, G. R. de Snoo, L. Rakosy,

and C. Ramwell. 2009. Ecological impacts of early 21st century agricultural change in

Europe: a review. Journal of Environmental Management 91:22-46.

Suárez-Seoane, S., P. E. Osborne, and J. Baudry. 2002. Responses of birds of different

biogeographic origins and habitat requirements to agricultural land abandonment in

northern Spain. Biological Conservation 105:333-344.

Sutcliffe, L. M., P. Batáry, U. Kormann, A. Báldi, L. V. Dicks, I. Herzon, D. Kleijn, P. Tryjanowski,

I. Apostolova, and R. Arlettaz. 2014. Harnessing the biodiversity value of Central and

Eastern European farmland. Diversity and Distributions, doi:10.1111/ddi.12288.

Sutcliffe, L. M., I. Paulini, G. Jones, R. Marggraf, and N. Page. 2013. Pastoral commons use in

Romania and the role of the Common Agricultural Policy. International Journal of the

Commons 7:58-72.

Swenson, J., and H. Andrén. 2005. A tale of two countries: large carnivore depredation and

compensation schemes in Sweden and Norway. In: R. Woodroffe, S. Thirgood, and A.

Rabinowitz (eds). People and wildlife: conflict or co-existence? Cambridge University

Press, Cambridge, pp 323-339.

Swenson, J. E., B. Dahle, H. Busk, O. Opseth, T. Johansen, A. Söderberg, K. Wallin, and G.

Cederlund. 2007. Predation on moose calves by European brown bears. Journal of Wildlife


Swenson, J. E., N. Gerstl, B. Dahle, and A. Zedrosser. 2000. Action plan for the conservation of

the brown bear in Europe (Ursus arctos). Council of Europe, Strasbourg, 68 pp.

Swenson, J. E., A. Jansson, R. Riig, and F. Sandegren. 1999. Bears and ants: myrmecophagy by

brown bears in central Scandinavia. Canadian Journal of Zoology 77:551-561.

Swenson, J. E., P. Wabakken, F. Sandegren, A. Bjärvall, R. Franzén, and A. Söderberg. 1995. The

near extinction and recovery of brown bears in Scandinavia in relation to the bear

management policies of Norway and Sweden. Wildlife Biology 1:11-25.

References

236

Taboada, A., D. J. Kotze, R. Tárrega, and J. M. Salgado. 2006. Traditional forest management: do

carabid beetles respond to human-created vegetation structures in an oak mosaic

landscape? Forest Ecology and Management 237:436-449.

Takeuchi, K. 2010. Rebuilding the relationship between people and nature: the Satoyama Initiative.

Ecological Research 25:891-897.

Terborgh, J., and J. A. Estes 2010. Trophic cascades: predators, prey, and the changing dynamics of

nature. Island Press, Washington DC, 488 pp.

Terborgh, J., L. Lopez, P. Nunez, M. Rao, G. Shahabuddin, G. Orihuela, M. Riveros, R. Ascanio,

G. H. Adler, and T. D. Lambert. 2001. Ecological meltdown in predator-free forest

fragments. Science 294:1923-1926.

Teşculă, N., and G. A. 2007. The history of the protected area. In: Mihai Eminescu Trust (ed). The

Management plan of the Breite Reserve. Mihai Eminescu Trust, Sighisoara, pp 51-55 (in

Romanian).

Theuerkauf, J., and S. Rouys. 2008. Habitat selection by ungulates in relation to predation risk by

wolves and humans in the Białowieża Forest, Poland. Forest Ecology and Management

256:1325-1332.

Thirgood, S., R. Woodroffe, and A. Rabinowitz 2005. The impact of human-wildlife conflict on

human lives and livelihoods. In: R. Woodroffe, S. Thirgood, and A. Rabinowitz (eds).

People and wildlife: conflict or coexistence? Cambridge University Press, Cambridge, pp

13-26.

Thompson, G. G., and P. C. Withers. 2003. Effect of species richness and relative abundance on

the shape of the species accumulation curve. Austral Ecology 28:355-360.

Tilman, D., J. Fargione, B. Wolff, C. D'Antonio, A. Dobson, R. Howarth, D. Schindler, W. H.

Schlesinger, D. Simberloff, and D. Swackhamer. 2001. Forecasting agriculturally driven

global environmental change. Science 292:281-284.

Tilman D, K.G. Cassman, P. A. Matson, R. Naylor, S. Polasky. 2002. Agricultural sustainability and

intensive production practices. Nature 418(6898):671-677.

Tjernberg, M., K. Johnsson, and S. Nilsson. 1993. Density variation and breeding success of the

black woodpecker Dryocopus martius in relation to forest fragmentation. Ornis Fennica

70:155-155.

Torres, R. T., E. Virgós, J. Santos, J. D. Linnell, and C. Fonseca. 2012. Habitat use by sympatric red

and roe deer in a Mediterranean ecosystem. Animal Biology 62:351-366.

Treves, A., R. R. Jurewicz, L. Naughton-Treves, R. A. Rose, R. C. Willging, and A. P. Wydeven.

2002. Wolf depredation on domestic animals in Wisconsin, 1976-2000. Wildlife Society

Bulletin 30:231-241.

Treves, A., and K. U. Karanth. 2003. Human‐carnivore conflict and perspectives on xarnivore

management worldwide. Conservation Biology 17:1491-1499.

Treves, A., R. B. Wallace, L. Naughton-Treves, and A. Morales. 2006. Co-managing human–

wildlife conflicts: a review. Human Dimensions of Wildlife 11:383-396.

Tryjanowski, P., T. Hartel, A. Báldi, P. Szymański, M. Tobolka, I. Herzon, A. Goławski, M.

Konvička, M. Hromada, L. Jerzak, K. Kujawa, M. Lenda, G. Orłowski, M. Panek, P.

Skórka, T. H. Sparks, S. Tworek, A. Wuczyński, and M. Żmihorski. 2011. Conservation of

farmland birds faces different challenges in Western and Central-Eastern Europe. Acta

Ornithologica 46:1-12.

Tscharntke, T., A. M. Klein, A. Kruess, I. Steffan-Dewenter, and C. Thies. 2005. Landscape

perspectives on agricultural intensification and biodiversity: ecosystem service

management. Ecology Letters 8:857-874.

References

237

Tscharntke, T., J. M. Tylianakis, T. A. Rand, R. K. Didham, L. Fahrig, P. Batary, J. Bengtsson, Y.

Clough, T. O. Crist, and C. F. Dormann. 2012. Landscape moderation of biodiversity

patterns and processes: eight hypotheses. Biological Reviews 87:661-685.

Tucker, G. M., and M. I. Evans 1997. Habitats for birds in Europe: a conservation strategy for the

wider environment. Birdlife International, Cambridge, 464 pp.

Tyler, G. A., R. E. Green, and C. Casey. 1998. Survival and behaviour of Corncrake Crex crex chicks

during the mowing of agricultural grassland. Bird Study 45:35-50.

Uchida, K., and A. Ushimaru. 2014. Biodiversity declines due to abandonment and intensification

of agricultural lands: patterns and mechanisms. Ecological Monographs 84:637-658.

Uezu, A., J. P. Metzger, and J. M. Vielliard. 2005. Effects of structural and functional connectivity

and patch size on the abundance of seven Atlantic Forest bird species. Biological


Uğurlu, E., J. Roleček, and E. Bergmeier. 2012. Oak woodland vegetation of Turkey: a first

overview based on multivariate statistics. Applied Vegetation Science 15:590-608.

Van Uytvanck, J., D. Maes, D. Vandenhaute, and M. Hoffmann. 2008. Restoration of woodpasture

on former agricultural land: The importance of safe sites and time gaps before grazing for

tree seedlings. Biological Conservation 141:78-88.

Van Weperen, M. 2009. Habitat selection of the Corncrake (Crex crex) in floodplains along the

Dutch Rhine River branches. MSc thesis, Radboud Universiteit Nijmegen, 65 pp.

Vanak, A. T., C. R. Dickman, E. A. Silva-Rodriguez, J. R. Butler, and E. G. Ritchie. 2013. Top-dogs

and under-dogs: competition between dogs and sympatric carnivores. In: M. E. Gompper

(ed). Free-Ranging Dogs and Wildlife Conservation. Oxford University Press, Oxford, pp

69-93.

Vera, F. W. M. 2000. Grazing ecology and forest history. CAB International, Wallingford, 506 pp.

Verhulst, J., A. Báldi, and D. Kleijn. 2004. Relationship between land-use intensity and species

richness and abundance of birds in Hungary. Agriculture, Ecosystems & Environment

104:465-473.

Vickery, J., J. Tallowin, R. Feber, E. Asteraki, P. Atkinson, R. Fuller, and V. Brown. 2001. The

management of lowland neutral grasslands in Britain: effects of agricultural practices on

birds and their food resources. Journal of Applied Ecology 38:647-664.

Vitousek, P. M., H. A. Mooney, J. Lubchenco, and J. M. Melillo. 1997. Human domination of

Earth's ecosystems. Science 277:494-499.

Vojta, J., and L. Drhovská. 2012. Are abandoned wooded pastures suitable refugia for forest

species? Journal of Vegetation Science 23:880-891.

Voříšek, P., F. Jiguet, A. van Strien, J. Škorpilová, A. Klvaňová, and R. Gregory. 2010. Trends in

abundance and biomass of widespread European farmland birds: how much have we lost.

BOU Proceedings–Lowland Farmland Birds III, 24 pp.

Wegener, U. 1998. Naturschutz in der Kulturlandschaft: Schutz und Pflege von Lebensräumen.

Gustav Fischer, Jena, 456 pp (in German.).

Weibull, A.-C., Ö. Östman, and Å. Granqvist. 2003. Species richness in agroecosystems: the effect

of landscape, habitat and farm management. Biodiversity & Conservation 12:1335-1355.

Weibull, A. C., J. Bengtsson, and E. Nohlgren. 2000. Diversity of butterflies in the agricultural

landscape: the role of farming system and landscape heterogeneity. Ecography 23:743-750.

Wettstein, W., T. Szép, and M. Kéry. 2001. Habitat selection of Corncrakes (Crex crex L.) in

Szatmár-Bereg (Hungary) and implications for further monitoring. Ornis Hungarica 11:9-

18.

Whittingham, M. J., J. R. Krebs, R. D. Swetnam, J. A. Vickery, J. D. Wilson, and R. P. Freckleton.

2007. Should conservation strategies consider spatial generality? Farmland birds show

regional not national patterns of habitat association. Ecology Letters 10:25-35.

References

238

Widerberg, M. K., T. Ranius, I. Drobyshev, U. Nilsson, and M. Lindbladh. 2012. Increased

openness around retained oaks increases species richness of saproxylic beetles. Biodiversity

and Conservation 21:3035-3059.

Wiktander, U., I. Nilsson, S. Nilsson, and O. Olsson. 1992. Occurrence of the lesser spotted

woodpecker Dendrocopos minor in relation to area of deciduous forest. Ornis Fennica 69:113-

113.

Wiktander, U., O. Olsson, and S. G. Nilsson. 2001. Seasonal variation in home-range size, and

habitat area requirement of the lesser spotted woodpecker (Dendrocopos minor) in southern

Sweden. Biological Conservation 100:387-395.

Wilkie, K. 2001. The Saxon villages of Transylvania, Romania: a future for mediaeval landscape.

Report to HRH The Prince of Wales and Mihai Eminescu Trust, United Kingdom, 33 pp.

Wilkinson, N. I., J. D. Wilson, and G. Q. Anderson. 2012. Agri-environment management for

corncrake Crex crex delivers higher species richness and abundance across other taxonomic

groups. Agriculture, Ecosystems & Environment 155:27-34.

Williams, D. R., and S. I. Stewart. 1998. Sense of place: an elusive concept that is finding a home in

ecosystem management. Journal of Forestry 96:18-23.

Wilmers, C. C., R. L. Crabtree, D. W. Smith, K. M. Murphy, and W. M. Getz. 2003. Trophic

facilitation by introduced top predators: grey wolf subsidies to scavengers in Yellowstone

National Park. Journal of Animal Ecology 72:909-916.

Wilson, J. B., R. K. Peet, J. Dengler, and M. Partel. 2012. Plant species richness: the world records.

Journal of Vegetation Science 23:796-802.

Woodroffe, R. 2000. Predators and people: using human densities to interpret declines of large

carnivores. Animal Conservation 3:165-173.

Woodroffe, R., S. Thirgood, and A. Rabinowitz 2005. People and wildlife: conflict or co-existence?

Cambridge University Press, Cambridge, 516 pp.

Wretenberg, J., Å. Lindstrom, S. Svensson, and T. Pärt. 2007. Linking agricultural policies to

population trends of Swedish farmland birds in different agricultural regions. Journal of

Applied Ecology 44:933-941.

Wretenberg, J., T. Pärt, and Å. Berg. 2010. Changes in local species richness of farmland birds in

relation to land-use changes and landscape structure. Biological Conservation 143:375-381.

Wuczynski, A., K. Kujawa, Z. Dajdok, and W. Grzesiak. 2011. Species richness and composition of

bird communities in various field margins of Poland. Agriculture Ecosystems &


Young, J., A. Watt, P. Nowicki, D. Alard, J. Clitherow, K. Henle, R. Johnson, E. Laczko, D.

McCracken, S. Matouch, J. Niemela, and C. Richards. 2005. Towards sustainable land use:

identifying and managing the conflicts between human activities and biodiversity

conservation in Europe. Biodiversity and Conservation 14:1641-1661.

Young, J. C., M. Marzano, R. M. White, D. I. McCracken, S. M. Redpath, D. N. Carss, C. P. Quine,

and A. D. Watt. 2010. The emergence of biodiversity conflicts from biodiversity impacts:

characteristics and management strategies. Biodiversity and Conservation 19:3973-3990.

Zedrosser, A., B. Dahle, J. E. Swenson, and N. Gerstl. 2001. Status and management of the brown

bear in Europe. Ursus 12:9-20.

Zedrosser, A., S. M. J. G. Steyaert, H. Gossow, and J. E. Swenson. 2011. Brown bear conservation

and the ghost of persecution past. Biological Conservation 144:2163-2170.

Zinn, H. C., M. J. Manfredo, and D. J. Decker. 2008. Human conditioning to wildlife: steps toward

theory and research. Human Dimensions of Wildlife 13:388-399.

Zinn, H. C., M. J. Manfredo, and J. J. Vaske. 2000. Social psychological bases for stakeholder

acceptance capacity. Human Dimensions of Wildlife 5:20-33.

Zuur, A., E.N. Ieno, N. Walker, A. A. Saveliev, G. M. Smith. 2009. Mixed effects models and

extensions in ecology with R. Springer, New York, 574 pp.

239

Appendix I

240

241

Appendix I

Low-intensity agricultural landscapes in Transylvania support

high butterfly diversity: implications for conservation

Jacqueline Loos, Ine Dorresteijn, Jan Hanspach, Pascal Fust, László Rakosy, Joern Fischer

PLoS ONE (2014), 9:e103256

Appendix I

242

Butterfly diversity in traditional farming landscapes

243

Abstract

European farmland biodiversity is declining due to land-use changes towards agricultural

intensification or abandonment. Some Eastern European farming systems have sustained

traditional forms of use, resulting in high levels of biodiversity. However, global markets and

international policies now imply rapid and major changes to these systems. To effectively protect

farmland biodiversity, understanding landscape features which underpin species diversity is crucial.

Focusing on butterflies, we addressed this question for a cultural-historic landscape in Southern

Transylvania, Romania. Following a natural experiment, we randomly selected 120 survey sites in

farmland, 60 each in grassland and arable land. We surveyed butterfly species richness and

abundance by walking transects with four repeats in summer 2012. We analysed species

composition using Detrended Correspondence Analysis. We modelled species richness, richness of

functional groups, and abundance of selected species in response to topography, woody vegetation

cover and heterogeneity at three spatial scales, using generalised linear mixed effects models.

Species composition widely overlapped in grassland and arable land. Composition changed along

gradients of heterogeneity at local and context scales, and of woody vegetation cover at context and

landscape scales. The effect of local heterogeneity on species richness was positive in arable land,

but negative in grassland. Plant species richness, and structural and topographic conditions at

multiple scales explained species richness, richness of functional groups and species abundances.

Our study revealed high conservation value of both grassland and arable land in low-intensity

Eastern European farmland. Besides grassland, also heterogeneous arable land provides important

habitat for butterflies. While butterfly diversity in arable land benefits from heterogeneity by small-

scale structures, grasslands should be protected from fragmentation to provide sufficiently large

areas for butterflies. These findings have important implications for EU agricultural and

conservation policy. Most importantly, conservation management needs to consider entire

landscapes, and implement appropriate measures at multiple spatial scales.

Appendix I

244

Introduction

Almost half of Europe’s terrestrial surface consists of farmland, and many species, including rare

and endangered ones, depend on farmland as habitat (Stoate et al. 2009; Kleijn et al. 2011). The

loss of cultural-historic landscapes through intensification or abandonment of farming practices is

causing declines of farmland biodiversity (Benton et al. 2003; Thomas et al. 2004; Cremene et al.

2005; Foley et al. 2005; Stoate et al. 2009). To effectively design conservation strategies, knowledge

is needed about which variables influence species richness and distribution at different spatial scales

(Weibull et al. 2003; Kumar et al. 2009; Brückmann et al. 2010).

In Western Europe, species loss in farmland has been associated with an increase of

agricultural productivity (Maes & Van Dyck 2001; Weibull et al. 2003; Van Dyck et al. 2009), most

likely caused by the use of agrochemicals (McLaughlin & Mineau 1995) and the loss and

fragmentation of semi-natural patches, especially grasslands (Bergman et al. 2004; Brückmann et al.

2010). In Eastern Europe, socio-economic conditions and land-use have been rapidly changing

since the breakdown of communism and accession of new member states to the European Union

(EU) (Kuemmerle et al. 2008; Kluvánková-Oravská et al. 2009; Mikulcak et al. 2013). Current

changes involve a dual threat to biodiversity, with a trend towards structural simplification on the

one hand and abandonment of low-intensity practices on the other hand (Schmitt & Rákosy 2007;

Young et al. 2007). The current situation in Eastern Europe thus differs in important ways from

Western European countries (Pullin et al. 2009; Stoate et al. 2009; Tryjanowski et al. 2011), and a

better understanding is needed of how organisms respond to landscape features within low-

intensity farming areas of Eastern Europe.

Heterogeneous landscapes typically harbour greater species richness than homogenous

landscapes (Benton et al. 2003; Tscharntke et al. 2005; Ekroos et al. 2013), most likely because of

their greater niche diversity, as well as spillover effects and habitat complementation (Dunning et

al. 1992). Agricultural simplification and land abandonment typically lead to a loss of landscape

connectivity, which may not only dissect the habitats for species, but also causes flow-on effects on

the composition and configuration of the landscape as a whole (Persson et al. 2010; Fahrig et al.

2011).

A particularly interesting cultural-historic region in Eastern Europe is Transylvania, which

supports extraordinarily high levels of farmland biodiversity (Fischer et al. 2012; Page et al. 2012).

Especially in its South, Transylvania is characterised by a small-scale mosaic of different low-

intensity land-uses that provide many different, well-connected structures such as field margins and

roadside vegetation. The historic management of the area has created heterogeneity at multiple

spatial scales: within tens of metres (hereafter termed the local scale), in the immediate

surroundings around any given location (the context scale), as well as over thousands of metres (the

landscape scale) (Akeroyd & Page 2006; Hartel et al. 2008).

Here, we focus on butterflies as a taxonomic group that rapidly responds to environmental

changes (Erhardt 1985) and is known to be sensitive to land-use change worldwide (Thomas et al.


245

2004). In Europe, many butterflies use anthropogenic landscape elements (van Swaay et al. 2006),

but species with different traits are expected to respond differently to land-use change (Krauss et al.

2003; Brückmann et al. 2010). For example, Öckinger & Smith (2006) found that the effects of

landscape composition differed between species of different mobility classes, and Börschig et al.

(2013) found that intensively used agricultural landscapes mostly support generalists. Yet, evidence

on the responses of butterflies to gradients of spatial heterogeneity is sparse, and more thorough

studies at multiple scales are needed (Öckinger et al. 2009; Ekroos et al. 2013).

We sought to understand the responses of butterfly diversity to key landscape gradients in

Southern Transylvania, using a snapshot natural experiment (Diamond 1986; Lindenmayer et al.

2008) that spanned the full range of environmental conditions with respect to heterogeneity and

woody vegetation cover across multiple scales. Our overarching aim was to understand drivers of

species richness and composition. Specifically, we asked (i) how landscape structures affect the

composition of butterfly communities; (ii) which landscape structures explain butterfly species

richness at various spatial scales; and (iii) which landscape structures affect abundance patterns of

selected species. We discuss our findings in the context of possible landscape changes that may take

place in Transylvania.

Materials and methods

Ethics Statement

We obtained the necessary permit for surveying butterflies within the EU Natura 2000 network

from Progresul Silvic, the organization officially entrusted with the custody of the protected area by

the Romanian government. The survey procedure was approved beforehand by the ethics

committee of Leuphana University Lueneburg.

Data Availability Statement

All data underlying the findings reported in this study are available from the Dryad Digital

Repository (http://doi.org/10.5061/dryad.97s1k).

Study area and experimental design

The study area covered approximately 7,000 km2 in the lowlands of Southern Transylvania,

Romania (Fig. A1.1). We followed the notion of a natural experiment (Diamond 1986), with

randomised site selection in pre-defined strata at two levels: study villages and survey sites within

villages.

To select study villages, we first allocated each raster pixel of the study area to different

‘village catchments’. These were calculated using a cost-distance algorithm in ARCGIS with the

village centre as the reference point and the slope and the distance to the next village as the cost

Appendix I

246

variables. Information about village locations was extracted from CORINE land cover data 2006

(http://www.eea.europa.eu/data-and-maps/data#c12=corine+land+cover+version+13&b_start=

0&c17=CLC2006), and slope was calculated from the digital elevation model ASTER (Advanced

Spaceborne Thermal Emission and Reflection Radiometer). Topographically based village

catchments were used instead of administrative boundaries because administrative boundaries were

only available at the commune level (typically 3-5 villages). However, we found that the resulting

polygons accurently reflected historical land-use responsibilities. Second, we stratified village

catchments along a gradient of terrain ruggedness and according to their protection status under

the EU Birds and Habitats Directives. Terrain ruggedness was calculated as the standard deviation

of the altitude of the catchment, and we used quantiles to classify ruggedness as either low, medium

or high. Protection status of the catchments was either unprotected, SCI (Site of Community

Importance) or SPA (Special Protection Area). Third, we randomly chose 30 villages, covering all

combination of ruggedness and protection status (Table SA1.1).

To select survey sites, we stratified the agricultural area within these 30 villages according

to CORINE land cover as grassland or arable land and excluded other land cover classes. Within

these strata, we spanned two gradients that we assumed sensitive to change in the future as a result

of structural simplification, namely woody vegetation cover and heterogeneity. We estimated

woody vegetation cover in a circular one hectare area based on classified 10 m SPOT data (CNES,

ISIS programme). To assess heterogeneity, we used the standard deviation of 2.5 m panchromatic

SPOT data within a one hectare circle. We assigned each hectare of the agricultural landscape to a

combination of three classes of woody vegetation cover by three classes of heterogeneity. We

distinguished low (0-5%), medium (>5-15%) and high (>15%) woody vegetation cover and used

the lower, middle and upper third of percentiles to classify heterogeneity. Within these

combinations, we randomly selected replicates for each cross-combination (except for the

combination of high heterogeneity and low woody vegetation cover, which did not exist (Table

SA1.2)). In total, we selected 120 circular 1 ha survey sites, with 60 in grassland and 60 in arable

land, and an average of four survey sites per village catchment. Notably, sites in arable land in this

context were consciously placed not to represent only arable fields specifically, but rather to

capture the whole range of conditions within the mosaic of arable land (Bennett et al. 2006),

including field margins and fallow land.


247

Figure A1.1 Location of the study area with investigated village catchments in Transylvania, Romania. The

small letters indicate the village catchments illustrated for predictions in Figure A1.4 (a= Cincu, b= Granari,

c= Viscri).

Data collection

Butterfly surveys (response data)

We assessed species richness and abundance of butterflies (Rhopalocera) and diurnal burnet moths

(Zygaenidae) by walking four transects of 50 m length per survey site (Pollard & Yates 1993). We

included burnet moths because they are comparable to butterflies in their ecology (Naumann et al.

1999; Öckinger & Smith 2006). These transect pointed north, east, south and west, and started 6 m

from the centre of a given site. In a given transect walk, each butterfly observed within 2.5 m of

each side of the transect and 5 m in front of the observer was identified and counted. Species that

we could not identify in the field were treated as compound species: L. sinapis/ juvernica, C.

alfacariensis/ hyale and Zygaena minos/ purpuralis. Adscita, Jordanita and Carcharodus occurred within the

study region, and are represented by two, two and three species, respectively (Rakosy et al. 2003).

However, these species are difficult to distinguish and therefore were only identified to the genus

level. Surveys were repeated on four occasions between May and August 2012 by four different,

trained observers. Surveys were conducted under suitable weather conditions (no rain, <90% cloud

cover, >17 °C, no strong wind), between 9 am and 5 pm.

Appendix I

248

Environmental data (explanatory variables)

We followed a multi-scale approach and included explanatory variables that could potentially

explain butterfly distribution at the local (1 ha), context (50 ha) and landscape scale (i.e. village

catchments, ranging from 430 to 4963 ha). An overview of all variables included in the analysis is

presented in Table A1.1.

At the local scale, we collected data on vascular plant species richness in eight randomized

quadrants (1x1 m). We used cumulative plant species richness per site as an explanatory variable.

We also calculated indices for heatload (after Parker 1991) and terrain wetness as a measure of

potential soil wetness, and included heterogeneity assessed by the spectral variance of SPOT data

(see Table A1 for details). We calculated percent woody vegetation cover at local and context

scales, and used CORINE land cover to calculate percent forest at the landscape scale. For the

context and landscape scales, we calculated the terrain ruggedness as the standard deviation of

altitude. We also quantified compositional or configurational heterogeneity of the different land

covers grassland, arable land and forest as provided by CORINE land cover data. At the context

scale, our chosen heterogeneity measures (Simpson index of land cover diversity, edge density)

were correlated (ρ = 0.76). Hence, we included only edge density as an explanatory variable

(following Kumar et al. 2009). At the landscape scale, we used both edge density and the Simpson

index of diversity and added the amount of pasture and forest per village catchment, based on

CORINE land cover data. Variables on compositional and configurational heterogeneity were

calculated using FRAGSTATS v4.2 (McGarigal et al. 2012) and all other variables using ARCGIS

10.1 (ESRI Inc., Redland, CA).


249

Table A1.1 Definition of environmental variables used in the study at three different scales and method of

obtaining those. Abbreviations are used in Figure A1.2 and Table A1.2.

Scale Variable (abbreviation) Definition and method

Local (1 ha)

Number of plants species (NoPlant)

Vascular plant species richness assessed by eight randomly distributed quadrants of one by one meter

Heterogeneity (het_1ha) Heterogeneity measured as the standard deviation of 2.5 m panchromatic SPOT picture (©CNES, ISIS programme)

Woody vegetation cover (woody_1ha)

Proportion of woody vegetation cover based on classified 10m SPOT satellite image (©CNES, ISIS programme)

Heat index (heatload) Potential for ground heating calculated after Parker (1991): Heat index = cos (slope aspect -225) * tan (slope angle)

Terrain Wetness Index (TWI) Measure of potential soil wetness, estimated as the position in the landscape and the slope from ASTER digital elevation model with 30m resolution.

Land Cover (LU_type) Land-use classification as arable land, grassland or forest based on CORINE land cover

Context (50 ha)

Ruggedness (rugg_50ha) Terrain ruggedness, calculated as standard deviation of altitude

Woody vegetation cover (woody_50ha)

Proportion of woody vegetation cover based on classified 10m SPOT satellite image

Configurational heterogeneity (ED_50ha)

Configuration of different land covers, calculated as the edge density with FRAGSTATS v4.2 based on CORINE land cover

Landscape (village catchment)

Amount of pasture (past_catch) Proportion of pasture, based on CORINE land cover

Woody vegetation cover (woody_catch)

Proportion of forest cover based on CORINE land cover

Ruggedness (catch_rugg) Terrain ruggedness, calculated as the standard deviation of the altitude

Compositional heterogeneity (SIDI)

Composition of different land covers, calculated as Simpson index of diversity with FRAGSTATS v4.2 based on CORINE land cover

Configurational heterogeneity (ED) Configuration of different land covers, calculated as edge density with FRAGSTATS v4.2 based on CORINE land cover

Random effects

Village catchment Classification of the landscape into social-ecological units according to a cost distance algorithm of proximity to the nearest village as reference point and the slope of the terrain as cost factor

Level Observation level random effect

Analysis

We pooled all observed butterfly species and individuals from the four survey rounds for each

survey site. First, we tested for differences in species richness and abundance between different

levels of official protection by using Analysis of Variance (ANOVA). Second, we conducted a

detrended correspondence analysis (DCA) to describe species composition and its relation to

environmental variables. We used a permutation test to fit and test the correlation of

environmental variables with the ordination.

Third, we used generalized linear mixed effects models (GLMMs) to assess effects of

environmental variables on butterfly species richness. Beforehand, we tested the explanatory

variables for collinearity (all ρ < 0.7; Table SA1.3; (Dormann et al. 2013)). We log-transformed

woody vegetation cover at local and context scales and heterogeneity at the local scale because

Appendix I

250

these variables were highly skewed. All numerical explanatory variables were scaled to mean zero

and unit variance. We included the variables listed in Table A1.1 to model species richness of

butterflies. To test for a unimodal relationship in response to woody vegetation cover, we included

a quadratic term of local woody vegetation cover. We furthermore expected that the effect of

heterogeneity may differ between grassland and arable land and therefore included an interaction

term between land cover type and heterogeneity. Grasslands are also interesting to look at

separately because they are among the most species rich biotopes for butterflies in Europe (van

Swaay et al. 2011). We assessed the variance inflation factor (VIF) of the generalized linear model

(GLM) and tested for spatial auto-correlation in the residuals. We included the village catchment as

a random effect and corrected for overdispersion by adding an observation level random effect. We

simplified the model by stepwise backwards selection retaining all variables with p<0.1. For

GLMMs, significance levels are only approximations, hence many statisticians suggest using a

significance level of p<0.1 (Bolker et al. 2009).

Likewise, we modelled species richness of functional groups. To this end, we distinguished

between species of low mobility (Bink´s mobility classes 1 and 2) and high mobility (Bink´s

mobility classes 7, 8 and 9; Bink 1992)). Highly mobile species were Colias crocea, Pieris brassicae,

Vanessa atalanta and Vanessa cardui. Low-mobility species were Brenthis daphne, Brenthis ino,

Coenonympha glycerion, Cupido minimus, Euphydryas aurinia, Hamaeris lucina, Heteropterus morpheus, Lopinga

achine, Melitaea britomartis, Melitaea diamina and Satyrium acaciae. As a third group we modelled the

richness of grassland specialists, namely Euphydryas aurinia, Polyommatus coridon, Cyaniris semiargus,

Lysandra bellargus, Phengaris arion, Cupido minimus and Erynnis tages (van Swaay et al. 2013).

We also modelled the abundance of individual species considered to be declining in

Western and Northern Europe, but that are widespread or even increasing in Eastern Europe (van

Swaay & Warren 1999; Konvicka et al. 2003; Franzén & Ranius 2004; van Swaay et al. 2013). We

only used species that were common enough in the study area to obtain reliable models, namely

Maniola jurtina, Coenonympha pamphilus, Polyommatus Icarus, Lycaena dispar and Glaucopsyche alexis. We

performed all statistical analyses in R (R Core Team 2013), using the packages MASS, ade4, vegan,

gdata and lme4.

Results

In total, we counted 19,878 individuals of 112 species of butterflies (Table SA1.4). Site-level species

richness varied between three and 45, and the number of individuals between seven and 452.

Eighty-five percent of all individuals belonged to 12 species: Colias alfacariensis/hyale, Minois dryas,

Aphantopus hyperantus, Pieris rapae, Everes argiades, Coenonympha glycerion, Leptidea sinapis/juvernica,

Melanargia galathea, Coenonympha pamphilus, Maniola jurtina, Polyommatus icarus, and Plebeius argus. SCI,

SPA and unprotected sites did not differ in species richness (F=0.54, p=0.58) but SCI sites

appeared to have a slightly lower abundance of individuals than unprotected sites (F=2.37, p=0.09).


251

Arable land and grassland did not differ in species richness (F=1.32, p=0.25) nor abundance of

individuals (F=1.51, p=0.22).

Multivariate analysis showed substantial overlap in species composition between arable

land and grassland (Fig. A1.2), with less than one complete species turnover (length of first axis =

2.9). The first axis (Eigenvalue = 0.21) described a gradient from sites with a low terrain wetness

index in homogenous landscapes to sites with a high terrain wetness index within highly

heterogeneous landscapes. The second DCA axis (Eigenvalue = 0.18) described a gradient from

low to high richness of vascular plants, ruggedness, woody vegetation cover and context-level

heterogeneity and landscape-level woody vegetation cover.

Figure A1.2 DCA ordination plot of butterfly species, with significant environmental variables superimposed

(p<0.05) (Abbreviations: NoPlant = Local plant species richness; TWI = Local terrain wetness index;

rugg_50ha = context terrain ruggedness; woody_50ha = context woody vegetation cover; ED_50ha =

context edge density; woody_catch = landscape woody vegetation cover; SIDI = landscape compositional

heterogeneity; Table A1.1).

Appendix I

252

Butterfly species richness was positively related to local plant species richness and local

woody vegetation cover, but negatively to local heat load (Table A1.2). It increased in response to

local heterogeneity in arable sites, but not in grasslands (Fig. A1.3). Species richness furthermore

increased with configurational heterogeneity and ruggedness at the context scale, but decreased

with landscape woody vegetation cover. The models show suitable areas for species of

conservation interest exist throughout village catchments, especially in large grassland areas and

boundary areas of arable land (Fig. A1.4).

Species richness of mobile butterflies was highest in arable land, and responded positively

at the landscape scale to both compositional heterogeneity and ruggedness. By contrast, richness of

low-mobility species was negatively related to landscape configurational heterogeneity, but

responded positively to local-scale plant species richness and context heterogeneity (for additional

details, see Table A1.2). Richness of specialist species was higher in grassland, in landscapes with

high terrain ruggedness and at sites with high plant species richness.

Figure A1.3 Predicted effect of local heterogeneity on species richness in arable land versus grassland, based

on the simplified generalized linear mixed model (Table A1.2).


253

Figure A1.4 Maps of predicted butterfly distributions in three example villages. Left: Land cover map

according to CORINE 2006; middle: predicted species richness for arable and grassland areas within each

village catchment; right: predicted abundance of the Meadow Brown (Maniola jurtina).

For individual species, both L. dispar and G. alexis were more abundant in arable land, and were

positively related to local plant species richness. L. dispar also responded positively to local woody

vegetation cover, but negatively to local heatload, whereas G. alexis showed a positive response to

context ruggedness and the amount of grassland in the landscape. The abundances of P. icarus, M.

jurtina and C. pamphilus increased with heterogeneity in arable land, but not in grassland, and

decreased with increasing heatload. Abundance of M. jurtina and C. pamphilus were positively related

to local plant species richness, and negatively to landscape woody vegetation cover. P. icarus

responded positively to the amount of grassland in the landscape. Abundance of C. pamphilus was

unimodally related to local woody vegetation cover.

Appendix I

254

Table A1.2 Parameter estimates of the species distribution models with significance levels indicated by: †

P<0.1; *P<0.05; **P<0.01; ***P<0.001. Arable land was used as the baseline land cover in all models. See

Table A1.1 for abbreviations.

Discussion

We found a high diversity of butterflies in the cultural-historic landscape of Southern Transylvania.

This is especially the case considering that we did not seek out sites expected a priori to harbour

great diversity, but rather surveyed randomly selected sites within the agricultural matrix. An even

greater diversity of butterflies, including rare and endangered species, would be expected to occur

in dry grassland patches and traditionally managed hay-meadows, which occur within our study

area but which we did not specifically target. Our findings suggest that some types of land-use

change could pose serious threats to butterfly diversity in Transylvania. Our findings can be

summarised within four themes, which we discuss in the following: (i) both grassland and arable

land have conservation value; (ii) low-intensity landscapes provide important resources for

butterflies; (iii) heterogeneity has a different effect in arable land than in grassland; and (iv) it is

important to consider multiple scales for effective butterfly conservation.

Species Richness

High mobile species

Low mobile species

Spec-ialists

L. dispar

G. alexis

P. icarus

M. jurtina

C. pam-philus

Intercept 3.026 -0.739 0.681 -0.436 0.581 -1.172 2.914 2.637 1.959

NoPlant 0.261 ***

0.600 ***

0.685 ***

0.581 *

0.941 **

0.597 ***

0.313 ***

LU_type -0.243 **

-1.250 ***

0.052 †

-2.059 ***

-1.554 *

-0.040 0.046 0.210

het_1ha 0.109 *

0.278 **

0.235 *

0.224 *

LU_type *het_1ha

-0.140 *

-0.326 *

-0.415 **

-0.387 *

TWI

woody_ 1ha

0.072 *

-0.054 0.443 †

0.102

woody_ 1ha^2

0.232 †

0.177*

heatload -0.057 *

-0.622 †

-0.167 *

-0.321 ***

-0.207 **

rugg_ 50ha

0.064 †

0.511 *

woody_ 50ha

ED_ 50ha

0.077 *

0.261 *

woody_ catch

-0.079 *

-0.412 ***

-0.232 †

past_ catch

0.721 **

0.256

rugg_ catch

-0.423 *

0.249 *

0.051 **

ED -0.374 **

SIDI 0.448 *


255

Both grassland and arable land have conservation value

Our findings revealed a high conservation value for butterflies of the small-scale farming system in

the lowlands of Transylvania. Interestingly, butterfly species richness and abundance were similar in

arable land and grassland. This is a surprising result and suggests a need to broaden the emphasis of

conservation activities from grassland protection towards the maintenance of heterogeneous

mosaic farmland, including cropland (Bennett et al. 2006). This is particularly important in the

context of criticisms that the recent reform of the European Union’s Common Agricultural Policy,

for example, falls far short of what is needed in terms of biodiversity conservation (Pe'er et al.

2014). Throughout Europe, grasslands are considered most important for butterfly conservation

(e.g. Brückmann et al. 2010; van Swaay et al. 2011). Arable land, on the other hand, has received far

less attention. In Western Europe, arable land has been found to support lower species richness

and more homogenous butterfly communities than grassland (Weibull & Ostman 2003; Weibull et

al. 2003). Our results indicate that this situation may be different in Eastern Europe, and that

certain types of arable land can in fact support similar levels of butterfly diversity as grasslands. A

possible explanation for the similar species richness in arable land and grassland in Transylvania

may be spillover effects (Dunning et al. 1992), which may be more likely in small-scale mosaics of

land covers. The mosaic character of the landscape also could explain the strong overlap in

butterfly communities between arable land and grassland.

Low-intensity landscapes provide important resources for butterflies

The fine-grained mosaic nature of arable land and the low-intensity nature of grassland in Southern

Transylvania emphasize that low-intensity land-use practices have major benefits for butterfly

conservation. Semi-natural elements occur throughout the landscape, and are a likely reason why

species richness is high throughout different land covers (Ekroos et al. 2013). Furthermore, species

richness of vascular plants can be high in field margins, which in turn may indicate high quality

habitat for butterflies (Steffan-Dewenter & Tscharntke 2000). Consistent with the findings of

Kumar, Simonson & Stohlgren (2009), we found plant species richness strongly related to butterfly

species richness. Currently, Transylvania contains some of the world’s most species rich areas for

plants (Wilson et al. 2012), which is partly linked to the low use of fertilizers (Jones 2009).

Agricultural intensification, by contrast, would likely lead to increased use of fertilizers and hence

reduced plant species richness (Zechmeister et al. 2003; Van Landuyt et al. 2008; Kleijn et al. 2009).

Furthermore, intensification is typically associated with the use of fewer, high yielding crop

varieties. Interestingly, many butterflies in Transylvania use the common crop Medicago sativa ssp.

sativa (Alfalfa), a leguminous species that provides nectar and that we also observed to serve as a

host plant for several butterfly species (e.g. Glaucopsyche alexis). Alfalfa is grown in small parcels, is

primarily used as winter fodder for livestock, and may easily be lost as a result of intensification.

However, high amounts of floral resources are critically important to maintain butterfly diversity.

Similarly, woody vegetation offers important resources for butterflies, including shelter and space

Appendix I

256

for thermoregulation (Dover et al. 1997). At present, Transylvania contains many scattered trees

and hedgerows, and we found that butterfly species richness responded positively to these

structures at the local scale. By contrast, a large amount of woody vegetation at the landscape scale

may lead to decreased species richness, probably due to a lack of open habitat.

Heterogeneity has a different effect in arable land than in grassland

We considered heterogeneity and woody vegetation cover at the local scale as two potentially

important gradients describing the structure of the landscape. Interestingly, our results showed that

the effect of local heterogeneity on species richness depended on land cover. In arable land, species

richness increased with heterogeneity, supporting our hypothesis that small-scale farming benefits

biodiversity by providing a range of different resources for butterflies. Notably, our land-use class

of ‘arable land’ reflected the highly heterogeneous nature of traditional farmland, and included

cropped areas as well as fallows and uncultivated field margins. These non-cropped areas are likely

to be particularly important to maintain butterfly diversity in arable land. By contrast, in grassland,

high heterogeneity was associated with reduced butterfly diversity. A possible explanation for this

pattern is that heterogeneity of grassland may correspond to a higher degree of fragmentation of

butterfly habitat, with likely negative consequences for species diversity (Krauss et al. 2004). Our

study thus confirms that heterogeneity per se is not universally beneficial for species richness (see

also Ekroos et al. 2008), although most work to date has focused on its positive effects (e.g. Kerr et

al. 2001).

The importance of considering multiple scales

To date, results from studies investigating multiple scales have been disparate and difficult to

generalize (Flick et al. 2012). We included three spatial scales in our study which we considered

relevant for butterfly diversity and distribution. Our study revealed that all investigated scales

affected butterfly community composition. Previous studies found local factors affecting butterfly

species composition, with local heterogeneity in land cover being a good predictor for species

composition in Canada (Kerr et al. 2001; Weibull & Ostman 2003). Butterfly species composition

in Transylvania also showed a significant correlation with local factors, but was explained by

heterogeneity and woody vegetation cover only at the two larger scales. Butterfly species richness

also responded to variables at all different spatial scales, especially at the local scale, but also at the

two larger scales (see also Weibull et al. 2003). This suggests that local habitat conditions are

particularly important, yet these cannot be considered in isolation from the surrounding landscape

(Steffan-Dewenter et al. 2002; Öckinger & Smith 2006).

Our models also showed that the different functional groups of butterflies were affected

by variables from different spatial scales. For example, landscape heterogeneity appeared to benefit

mobile species but not low-mobility species. Furthermore, we found that woody vegetation cover

was related to species richness. Land abandonment induces natural succession, whereas


257

intensification leads to loss of scattered woody vegetation, and both have negative effects on

butterfly richness in the long term (Baur et al. 2006). Both processes decrease structural

heterogeneity, which is important for viable butterfly populations in agricultural landscapes. In our

study, only Coenonympha pamphilus showed a unimodal relationship to local woody vegetation cover.

For such low-mobile species, presence of woody vegetation is crucial for wind shield and

thermoregulation. C. pamphilus is abundant in Transylvania, however its population state in other

European countries is declining (Conrad et al. 2007). Habitat heterogeneity from different spatial

scales, including the presence of woody vegetation, should be further investigated as possible key

elements in landscapes to halt biodiversity loss in farmland.

Conclusion

Collapse of communism and accession of Romania to the European Union have accelerated land-

use change in the rural areas of Transylvania, in particular towards land abandonment and

agricultural intensification. The two key gradients considered in this study, namely woody

vegetation cover and heterogeneity, would fundamentally change as a result of these two land-use

change processes. Along the gradients of woody vegetation cover and heterogeneity, we were able

to show that butterfly abundance and distribution were affected by a range of different variables

operating at multiple spatial scales. Not only local conditions, but the composition and

configuration of the landscape as a whole need to be considered for effective conservation

management of butterflies in low-intensively managed farming landscapes such as in Transylvania.

Our results showed that, unlike in Western Europe, species richness of butterflies was not

only high in grassland, but also in arable land. This suggests that more emphasis needs to be placed

on low-intensity farming practices and management of the landscape mosaic, and that arable land

needs to be actively considered in butterfly conservation strategies. In our study area, butterfly

richness would likely benefit from (1) the continuation of small-scale farming; (2) the production of

a variety of crops, including legume species; and (3) the maintenance of broad field margins and

uncultivated ruderal areas. New payment schemes under the Common Agricultural Policy have

recently been criticised as grossly inadequate (Pe'er et al. 2014). Our findings suggest that even

measures considered adequate in Western Europe may not be directly transferable to Transylvania

– in low-intensity landscapes, it will be particularly important to consider the high nature value that

entire agro-ecosystems provide, both inside and outside of protected areas (see also González-

Estébanez et al. 2011). Ultimately, the continued existence of historic-cultural landscapes such as

those in Transylvania hinges on the successful transfer of its appreciation and historic management

to future generations of farmers. Substantial efforts are therefore needed in environmental

education and in developing alternative ways for local people to make a living, for example through

the development of cultural and ecological tourism.

Appendix I

258

Acknowledgements

We warmly thank Rémi Bigonneau, Paul Kirkland, Joerg Steiner, Kimberley Pope, and Elek Telek

for help in the field. Sincere thanks to the farmers, landowners, mayors and the Mihai Eminescu

Trust for their support. We are grateful for analytical support by Henrik von Wehrden and Dave

Abson. Comments by two anonymous reviewers greatly helped to improve an earlier version of

this manuscript.


259

References

Akeroyd, J. R., and J. N. Page. 2006. The Saxon villages of southern Transylvania: Conserving

biodiversity in a historic landscape. In: D. Gafta, and J. Akeroyd (eds). Nature

conservation: concepts and practice, Springer, New York, pp 199-210.

Baur, B., C. Cremene, G. Groza, L. Rakosy, A. A. Schileyko, A. Baur, P. Stoll, and A. Erhardt.

2006. Effects of abandonment of subalpine hay meadows on plant and invertebrate

diversity in Transylvania, Romania. Biological Conservation 132:261-273.

Bennett, A. F., J. Q. Radford, and A. Haslem. 2006. Properties of land mosaics: implications for

nature conservation in agricultural environments. Biological Conservation 133:250-264.


the key? Trends in Ecology & Evolution 18:182-188.

Bergman, K. O., J. Askling, O. Ekberg, H. Ignell, H. Wahlman, and P. Milberg. 2004. Landscape

effects on butterfly assemblages in an agricultural region. Ecography 27:619-628.

Bink, F. A. 1992. Ecologische atlas van de dagvlinders van Noordwest-Europa. Schuyt & Co,

Haarlem, 512 pp.

Bolker, B. M., M. E. Brooks, C. J. Clark, S. W. Geange, J. R. Poulsen, M. H. H. Stevens, and J. S. S.

White. 2009. Generalized linear mixed models: a practical guide for ecology and evolution.

Trends in Ecology & Evolution 24:127-135.

Börschig, C., A.-M. Klein, H. von Wehrden, and J. Krauss. 2013. Traits of butterfly communities

change from specialist to generalist characteristics with increasing land-use intensity. Basic

and Applied Ecology 14:547-554.

Brückmann, S. V., J. Krauss, and I. Steffan-Dewenter. 2010. Butterfly and plant specialists suffer

from reduced connectivity in fragmented landscapes. Journal of Applied Ecology 47:799-

809.

Conrad, K. F., R. Fox, and I. P. Woiwod. 2007. Monitoring biodiversity: measuring long-term

changes in insect abundance. In: A. J. A. Stewart, T. R. New, and O. T. Lewis (eds). Insect

Conservation Biology. CAB International, Wallingford, pp 203-225.




Diamond, J. M. 1986. Overview: laboratory experiments, field experiments, and natural

experiments. In: J. M. Diamond, and T. J. Case (eds). Community Ecology. Harper & Row,

New York, pp 3-22.

Dormann, C. F., J. Elith, S. Bacher, C. Buchmann, G. Carl, G. Carre, J. R. G. Marquez, B. Gruber,

B. Lafourcade, P. J. Leitao, T. Munkemuller, C. McClean, P. E. Osborne, B. Reineking, B.

Schroder, A. K. Skidmore, D. Zurell, and S. Lautenbach. 2013. Collinearity: a review of

methods to deal with it and a simulation study evaluating their performance. Ecography

36:27-46.

Dover, J., T. Sparks, and J. Greatorex-Davies. 1997. The importance of shelter for butterflies in

open landscapes. Journal of Insect Conservation 1:89-97.

Dunning, J. B., B. J. Danielson, and H. R. Pulliam. 1992. Ecological processes that affect

populations in complex landscapes. Oikos 65:169-175.

Ekroos, J., M. Piha, and J. Tiainen. 2008. Role of organic and conventional field boundaries on

boreal bumblebees and butterflies. Agriculture, Ecosystems & Environment 124:155-159.

Ekroos, J., M. Rundlöf, and H. G. Smith. 2013. Trait-dependent responses of flower-visiting

insects to distance to semi-natural grasslands and landscape heterogeneity. Landscape

Ecology 28:1283-1292.

Appendix I

260

Erhardt, A. 1985. Diurnal lepidoptera: sensitive indicators of cultivated and abandoned grassland.

Journal of Applied Ecology 22:849-861.

Fahrig, L., J. Baudry, L. Brotons, F. G. Burel, T. O. Crist, R. J. Fuller, C. Sirami, G. M. Siriwardena,

and J. L. Martin. 2011. Functional landscape heterogeneity and animal biodiversity in

agricultural landscapes. Ecology Letters 14:101-112.


landscapes. Conservation Letters 5:167-175.

Flick, T., S. Feagan, and L. Fahrig. 2012. Effects of landscape structure on butterfly species

richness and abundance in agricultural landscapes in eastern Ontario, Canada. Agriculture


Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R. Carpenter, F. S. Chapin, M. T.

Coe, G. C. Daily, H. K. Gibbs, J. H. Helkowski, T. Holloway, E. A. Howard, C. J.

Kucharik, C. Monfreda, J. A. Patz, I. C. Prentice, N. Ramankutty, and P. K. Snyder. 2005.

Global consequences of land use. Science 309:570-574.

Franzén, M., and T. Ranius. 2004. Occurrence patterns of butterflies (Rhopalocera) in semi-natural

pastures in southeastern Sweden. Journal for Nature Conservation 12:121-135.

González-Estébanez, F. J., S. García-Tejero, P. Mateo-Tomás, and P. P. Olea. 2011. Effects of

irrigation and landscape heterogeneity on butterfly diversity in Mediterranean farmlands.


Hartel, T., C. I. Moga, K. Öllerer, I. Sas, L. Demeter, D. Rusti, and A. Balog. 2008. A proposal

towards the incorporation of spatial heterogeneity into animal distribution studies in

Romanian landscapes. North-Western Journal of Zoology 4:67-74.

Jones, A. 2009. Wildflower species indicators for lowland grassland habitat conservation in

Transylvania (Romania). Contributii Botanice 44:57-66.

Kerr, J. T., T. R. E. Southwood, and J. Cihlar. 2001. Remotely sensed habitat diversity predicts

butterfly species richness and community similarity in Canada. Proceedings of the National

Academy of Sciences 98:11365-11370.





Kleijn, D., M. Rundlöf, J. Scheper, H. G. Smith, and T. Tscharntke. 2011. Does conservation on


26:474-481.

Kluvánková-Oravská, T., V. Chobotová, I. Banaszak, L. Slavikova, and S. Trifunovova. 2009. From

government to governance for biodiversity: the perspective of central and Eastern

European transition countries. Environmental Policy & Governance 19:186-196.

Konvicka, M., M. Maradova, J. Benes, Z. Fric, and P. Kepka. 2003. Uphill shifts in distribution of

butterflies in the Czech Republic: effects of changing climate detected on a regional scale.

Global Ecology and Biogeography 12:403-410.

Krauss, J., A. M. Klein, I. Steffan-Dewenter, and T. Tscharntke. 2004. Effects of habitat area,

isolation, and landscape diversity on plant species richness of calcareous grasslands.

Biodiversity and Conservation 13:1427-1439.

Krauss, J., I. Steffan-Dewenter, and T. Tscharntke. 2003. How does landscape context contribute

to effects of habitat fragmentation on diversity and population density of butterflies?

Journal of Biogeography 30:889-900.

Kuemmerle, T., D. Muller, P. Griffiths, and M. Rusu. 2008. Land use change in Southern Romania

after the collapse of socialism. Regional Environmental Change 9:1-12.


261

Kumar, S., S. E. Simonson, and T. J. Stohlgren. 2009. Effects of spatial heterogeneity on butterfly

species richness in Rocky Mountain National Park, CO, USA. Biodiversity and


Lindenmayer, D. B., R. B. Cunningham, C. MacGregor, M. Crane, D. Michael, J. Fischer, R.

Montague-Drake, A. Felton, and A. Manning. 2008. Temporal changes in vertebrates

during landscape transformation: a large-scale “natural experiment”. Ecological

Monographs 78:567-590.

Maes, D., and H. Van Dyck. 2001. Butterfly diversity loss in Flanders (north Belgium): Europe's

worst case scenario? Biological Conservation 99:263-276.

McGarigal, K., S. Cushman, and E. Ene. 2012. FRAGSTATS v4: spatial pattern analysis program

for categorical and continuous maps. Software, University of Massachusetts, Amherst.

www.umass.edu/landeco/research/fragstats/fragstats.html.

McLaughlin, A., and P. Mineau. 1995. The impact of agricultural practices on biodiversity.



and biodiversity conservation in Central Romania. Environmental Conservation 40:129-

137.

Naumann, C. M., G. M. Tarmann, and W. G. Tremewan 1999. The Western Palaearctic Zygaenidae

(Lepidoptera). Apollo Books, Stenstrup, 304 pp.

Öckinger, E., M. Franzen, M. Rundlof, and H. G. Smith. 2009. Mobility-dependent effects on

species richness in fragmented landscapes. Basic and Applied Ecology 10:573-578.

Öckinger, E., and H. G. Smith. 2006. Landscape composition and habitat area affects butterfly

species richness in semi-natural grasslands. Oecologia 149:526-534.

Page, N., A. Bălan, S. Huband, R. Popa, L. Rákosy, and L. Sutcliffe. 2012. Romania. In: R.

Oppermann, G. Beaufoy, and G. Jones (eds). High Nature Value Farming in Europe.

Regionalkultur, Ubstadt-Weiher, pp 346-358.

Parker, K. C. 1991. Topography, substrate, and vegetation patterns in the Northern Sonoran

Desert. Journal of Biogeography 18:151-163.

Pe'er, G., L. V. Dicks, P. Visconti, R. Arlettaz, A. Báldi, T. G. Benton, S. Collins, M. Dieterich, R.

D. Gregory, F. Hartig, K. Henle, P. R. Hobson, D. Kleijn, R. K. Neumann, T. Robijns, J.

Schmidt, A. Shwartz, W. J. Sutherland, A. Turbé, F. Wulf, and A. V. Scott. 2014. EU

agricultural reform fails on biodiversity. Science 344:1090-1092.

Persson, A. S., O. Olsson, M. Rundlöf, and H. G. Smith. 2010. Land use intensity and landscape

complexity: analysis of landscape characteristics in an agricultural region in Southern

Sweden. Agriculture Ecosystems & Environment 136:169-176.

Pollard, E., and T. J. Yates 1993. Monitoring butterflies for ecology and conservation: the British

butterfly monitoring scheme. Chapman & Hall, London, 274 pp.

Pullin, A. S., A. Baldi, O. E. Can, M. Dieterich, V. Kati, B. Livoreil, G. Lovei, B. Mihok, O. Nevin,

N. Selva, and I. Sousa-Pinto. 2009. Conservation focus on Europe: major conservation

policy issues that need to be informed by conservation science. Conservation Biology

23:818-824.

R Core Team. 2013. R: a language and environment for statistical computing. R Foundation for

Statistical Computing, Vienna, Austria.

Rakosy, L., M. Goia, and Z. Kovács. 2003. Catalogul Lepidopterelor Romaniei/Verzeichnis der

Schmetterlinge Rumäniens. S. L. Romana, Cluj-Napoca, 79 pp.

Schmitt, T., and L. Rákosy. 2007. Changes of traditional agrarian landscapes and their conservation

implications: a case study of butterflies in Romania. Diversity and Distributions 13:855-

862.

Appendix I

262

Steffan-Dewenter, I., U. Munzenberg, C. Burger, C. Thies, and T. Tscharntke. 2002. Scale-

dependent effects of landscape context on three pollinator guilds. Ecology 83:1421-1432.

Steffan-Dewenter, I., and T. Tscharntke. 2000. Butterfly community structure in fragmented

habitats. Ecology Letters 3:449-456.

Stoate, C., A. Baldi, P. Beja, N. D. Boatman, I. Herzon, A. van Doorn, G. R. de Snoo, L. Rakosy,

and C. Ramwell. 2009. Ecological impacts of early 21st century agricultural change in

Europe: a review. Journal of Environmental Management 91:22-46.

Thomas, J. A., M. G. Telfer, D. B. Roy, C. D. Preston, J. J. D. Greenwood, J. Asher, R. Fox, R. T.

Clarke, and J. H. Lawton. 2004. Comparative losses of British butterflies, birds, and plants

and the global extinction crisis. Science 303:1879-1881.

Tryjanowski, P., T. Hartel, A. Baldi, P. Szymanski, M. Tobolka, I. Herzon, A. Golawski, M.

Konvicka, M. Hromada, L. Jerzak, K. Kujawa, M. Lenda, G. Orlowski, M. Panek, P.

Skorka, T. H. Sparks, S. Tworek, A. Wuczynski, and M. Zmihorski. 2011. Conservation of

farmland birds faces different challenges in Western and Central-Eastern Europe. Acta

Ornithologica 46:1-12.




Van Dyck, H., A. J. Van Strien, D. Maes, and C. A. M. Van Swaay. 2009. Declines in common,

widespread butterflies in a landscape under intense human use. Conservation Biology

23:957-965.

Van Landuyt, W., L. Vanhecke, I. Hoste, F. Hendrickx, and D. Bauwens. 2008. Changes in the

distribution area of vascular plants in Flanders (northern Belgium): eutrophication as a

major driving force. Biodiversity and Conservation 17:3045-3060.

van Swaay, C., D. Maes, S. Collins, M. L. Munguira, M. Sasic, J. Settele, R. Verovnik, M. Warren,

M. Wiemers, I. Wynhoff, and A. Cuttelod. 2011. Applying IUCN criteria to invertebrates:

How red is the Red List of European butterflies? Biological Conservation 144:470-478.

van Swaay, C., A. v. Strien, A. Harpke, B. Fontaine, C. Stefanescu, D. Roy, D. Maes, E. Kühn, E.

Õunap, E. Regan, G. Švitra, I. Prokofev, J. Heliölä, J. Settele, L. Pettersson, M. Botham,

M. Musche, N. Titeux, N. Cornish, P. Leopold, R. Julliard, R. Verovnik, S. Öberg, S.

Popov, S. Collins, S. Goloshchapova, T. Brereton, and M. Warren. 2013. The european

grassland butterfly indicator: 1990-2011. EEA Technical report. European Environment

Agency, Copenhagen, 36 pp.

van Swaay, C., M. Warren, and G. Lois. 2006. Biotope use and trends of European butterflies.

Journal of Insect Conservation 10:305-306.

van Swaay, C., and M. S. Warren 1999. Red data book of European butterflies (Rhopalocera).

Council of Europe, Strasbourg, 260 pp.

Weibull, A. C., and O. Ostman. 2003. Species composition in agroecosystems: the effect of

landscape, habitat, and farm management. Basic and Applied Ecology 4:349-361.

Weibull, A. C., O. Ostman, and A. Granqvist. 2003. Species richness in agroecosystems: the effect

of landscape, habitat and farm management. Biodiversity and Conservation 12:1335-1355.



Young, J., C. Richards, A. Fischer, L. Halada, T. Kull, A. Kuzniar, U. Tartes, Y. Uzunov, and A.

Watt. 2007. Conflicts between biodiversity conservation and human activities in the central

and eastern European countries. Ambio 36:545-550.

Zechmeister, H. G., I. Schmitzberger, B. Steurer, J. Peterseil, and T. Wrbka. 2003. The influence of

land-use practices and economics on plant species richness in meadows. Biological



263

Supporting Information for Appendix I

Table SA1.1 Number of the 30 focal villages within different strata. TRI= Terrain Ruggedness index,

Protection status is according to the EU Habitats and Birds Directives

Table SA1.2 Number of survey sites along the two gradients local heterogeneity and local woody vegetation

cover. A = Arable land and G = grassland Heterogeneity

Low Medium High

A G A G A G

Woody vegetation cover Low 8 7 7 8 0 0

Medium 7 8 8 7 7 8

High 8 7 7 8 8 7

Table SA1.3 Correlation matrix of the variables used in the study

Protection status

No SCI SPA

TRI Low 4 1 4

Medium 3 4 3

High 4 4 3

NoPlant het_

1ha

woody_

1ha

heat-

load

TWI rugg_

50ha

woody_

50ha

ed_

50ha

past_

catch

woody_

catch

catch_

rugg

SIDI ED

NoPlant 1.00 -0.22 0.07 0.03 -0.25 0.39 0.27 0.06 -0.09 0.14 -0.02 -0.17 -0.11

het_1ha -0.22 1.00 0.22 0.15 0.09 0.02 0.09 0.12 -0.18 0.15 0.03 0.02 0.11

woody_1ha 0.07 0.22 1.00 0.00 0.25 0.10 0.29 0.02 0.02 0.22 -0.04 -0.20 -0.15

heatload 0.03 0.15 0.00 1.00 -0.04 0.12 0.00 0.06 0.00 -0.01 0.13 0.15 0.18

TWI -0.25 0.09 0.25 -0.04 1.00 -0.13 -0.02 0.02 -0.05 0.17 0.10 0.04 -0.04

rugg_50ha 0.39 0.02 0.10 0.12 -0.13 1.00 0.49 0.27 -0.29 0.38 0.24 -0.07 0.15

woody_50ha 0.27 0.09 0.29 0.00 -0.02 0.49 1.00 0.42 -0.29 0.61 0.01 -0.27 0.14

ed_50ha 0.06 0.12 0.02 0.06 0.02 0.27 0.42 1.00 -0.29 0.30 0.20 -0.12 0.31

past_catch -0.09 -0.18 0.02 0.00 -0.05 -0.29 -0.29 -0.29 1.00 -0.46 -0.10 -0.11 -0.11

woody_catch 0.14 0.15 0.22 -0.01 0.17 0.38 0.61 0.30 -0.46 1.00 0.09 -0.29 -0.05

catch_rugg -0.02 0.03 -0.04 0.13 0.10 0.24 0.01 0.20 -0.10 0.09 1.00 0.25 0.09

SIDI -0.17 0.02 -0.20 0.15 0.04 -0.07 -0.27 -0.12 -0.11 -0.29 0.25 1.00 0.29

ED -0.11 0.11 -0.15 0.18 -0.04 0.15 0.14 0.31 -0.11 -0.05 0.09 0.29 1.00

Appendix I

264

Table SA1.4 Species list of butterfly species observed in the transects

Adscita sp Hipparchia fagi Polyommatus dorylas

Aglais urticae Inachis io Polyommatus icarus

Antocharis cardamines Iphiclides podalirius Polyommatus thersites

Apatura ilia Issoria lathonia Pontia edusa

Apatura iris Jordanita sp Pseudophilotes vicrama

Aphantopus hyperantus Lasiommata megera Pyrgus armoricanus

Aporia crataegi Leptidea morsei Pyrgus alveus

Araschnia levana Leptidea sinapis/ juvernica Pyrgus malvae

Argynnis adippe Limenitis camilla Rhagades pruni

Argynnis aglaja Limenitis populi Satyrium acaciae

Argynnis laodice Lopinga achine Satyrium ilicis

Argynnis niobe Lycaena alciphron Thymelicus lineola

Argynnis paphia Lycaena dispar Thymelicus sylvestris

Aricia agestis Lycaena phleas Vanessa atalanta

Aricia artaxerxes Lycaena thersamon Vanessa cardui

Boloria dia Lycaena tityrus Zygaena angelicae

Boloria euphrosyne Lycaena virgaureae Zygaena carniolica

Boloria selene Lysandra bellargus Zygaena ephialtes

Brenthis daphne Phengaris arion Zygaena filipendulae

Brenthis hecate Maniola jurtina Zygaena loti

Brenthis ino Melanargia galathea Zygaena minos/purpuralis

Brinthesia circe Meleagera daphnis Zygaena viciae

Cacharodus sp Melitaea athalia

Callophrys rubi Melitaea aurelia

Celastrina argiolus Melitaea britomartis

Coenonympha arcania Melitaea cinxia

Coenonympha glycerion Melitaea diamina

Coenonympha pamphilus Melitaea didyma

Colias alfacariensis/ hyale Melitaea phoebe

Colias crocea Melitaea trivia

Cupido decoloratus Minois dryas

Cupido minimus Neptis sappho

Cupido osiris Nymphalis antiopa

Cyaniris semiargus Ochlodes sylvanus

Erebia medusa Papilio machaon

Erynnis tages Pararge aegeria

Eumedonia eumedon Pieris brassicae

Euphydryas aurinia Pieris napi

Everes alcetas Pieris rapae

Everes argiades Plebejus argus

Glaucopsyche alexis Plebejus argyrognomon

Gonepteryx rhamni Plebejus idas

Hamearis lucina Polygonia c-album

Hesperia comma Polyommatus amandus

Heteropterus morpheus Polyommatus coridon

265

Appendix II

266

267

Appendix II

Plant diversity in a changing agricultural landscape mosaic in

Southern Transylvania (Romania)

Jacqueline Loos, Pavel Dan Turtureanu, Henrik von Wehrden, Jan Hanspach, Ine

Dorresteijn, József Pál Frink, Joern Fischer

Agriculture, Ecosystems & Environment (2014), 199:350-357

Appendix II

268

Plant diversity in traditional farming landscapes

269

Abstract

Traditional agricultural landscapes represent mosaics of land-use covers that often support high

species diversity. Many Eastern European countries contain large areas of High Nature Value

(HNV) farmland. However, these landscapes are likely to change under current EU regulations and

global market pressure, with potentially negative consequences for biodiversity. The conservation

value of Romania’s grasslands is widely recognized, but the potential conservation value of other

parts of the landscape mosaic has not been assessed to date. For this reason, we sought to assess

patterns of plant diversity across the entire landscape mosaic. We sampled vascular plants at 139

sites (comprising 8 plots of 1m2/ha) in forest (n = 23), grassland (n = 57) and within the arable

mosaic (n = 59). To examine potential differences in species richness and composition between

these land cover types, we used Analysis of Variance and Detrended Correspondence Analysis. We

also modeled total species richness, richness of HNV indicator plants and richness of arable weeds

in response to variables representing topography as well as structural and configurational

heterogeneity. Species composition differed strongly between grassland, the arable mosaic and

forests. Richness was highest in grasslands, but surprisingly, the arable mosaic and grassland

contributed similarly to the cumulative number of recorded species. Models of species richness

revealed a wide range of responses of plant groups to topographical conditions and to structural

and configurational heterogeneity, which often differed between land-use types. Plants were

affected by conditions measured at both local (1 ha) and landscape (50 ha) scales. Noting the

substantial, and hitherto under-recognised, contribution of the agricultural mosaic to regional-scale

plant diversity, we recommend consideration of the entire landscape mosaic in future conservation

schemes.

Appendix II

270

Introduction

Many agricultural landscapes around the world are characterized by a mosaic of land covers

(Forman 1995). With their various patches of land-use types and structures, mosaic landscapes

often host a wide range of species (Bennett et al. 2006). However, unprecedented changes in

agricultural mosaic landscapes are causing major biodiversity loss worldwide (Tscharntke et al.

2005). Moreover, patterns of species richness and distribution in agricultural landscapes are

affected by processes operating at multiple spatial scales, including both local and landscape-level

variables (Vandvik & Birks 2002; Rundlöf et al. 2010; Costanza et al. 2011). Hence, effective

management of biodiversity in agricultural landscapes requires an assessment of the drivers of

species diversity across multiple spatial scales.

In many Eastern European countries, such as Romania, traditional practices have created

small-scale mosaic landscapes. For example, 72% of farms in Romania are smaller than 1 ha

(Fundatia Adept 2012), and individual fields are typically smaller than that. However, Romania’s

farmland has been undergoing drastic changes since the collapse of communism in 1990

(Kuemmerle et al. 2008) and accession to the European Union (EU) in 2007. Ongoing land-use

changes comprise both intensification of land-use in some areas, and land abandonment in others

(Government of Romania 2010; Dahlström et al. 2013; Mikulcak et al. 2013). At present, twenty

percent of Romanian farmland is considered to be High Nature Value (HNV) farmland, and ten

percent is protected under the EU Birds and Habitats Directives (Natura 2000) (European

Environment Agency 2010). Despite official recognition of the ecological values of large areas of

farmland, the future of Romania’s agricultural landscapes and their biodiversity is uncertain.

Some of Romania’s most notable mosaic landscapes occur in the region of Southern

Transylvania. A large part of Transylvania was recently designated one of the largest continuous

(lowland) Natura 2000 sites in Europe (i.e. Târnavelor Plateau), partly in recognition of its

outstanding grassland diversity (Jones et al. 2010; Akeroyd & Page 2011). The region´s biodiversity

includes various taxa that are rare or endangered in other parts of Europe, such as the yellow-

bellied toad (Bombina variegata) (Hartel & von Wehrden 2013), Maculinea butterflies (Vodă et al.

2010) and several rare species of woodpeckers (Dorresteijn et al. 2013). Moreover, Transylvanian

dry grasslands hold the world record for vascular plant species richness at the scales of 0.1 m2 and

10 m2 (Wilson et al. 2012). However, land-use change is likely in Transylvania, and would pose

major threats to its biodiversity. Modifications of land-use will most likely consist of increasing

cropland area, increasing the use of agrochemicals, structural homogenization, and conversion of

traditional hay meadows to pastures, thus mirroring the patterns already apparent in Western

Europe (McLaughlin & Mineau 1995; Benton et al. 2003; Billeter et al. 2008; Ernoult & Alard

2011).

In this study we focus on vascular plants and their distribution throughout the entire

landscape mosaic in Southern Transylvania. Plants respond relatively slowly to environmental

changes (Helm et al. 2006), but in agricultural landscapes, specialized species are highly prone to


271

rapid decline (Davies et al. 2004; Clavel et al. 2010). Plant communities in agricultural landscapes

are at risk of homogenization in composition because of nutrient inputs, which many species, and

especially grassland specialists, are sensitive to. For Transylvania, a specific set of sensitive grassland

specialists has been proposed to indicate High Nature Value (HNV) grassland (Akeroyd & Bădărău

2012). Furthermore, arable weeds are of particular interest. Many such weeds have persisted in

Transylvania to date, but are under worldwide decline, and may react quickly to changes in the

environment, both at local and landscape scales (Gabriel et al. 2005; Armengot et al. 2012; Storkey

et al. 2012).

We sought to understand the responses of vascular plant diversity to key landscape

features. To that end, we used a snapshot natural experiment (Diamond 1986; Lindenmayer et al.

2008) that spanned a wide range of environmental conditions with respect to heterogeneity and

woody vegetation cover across local and landscape scales. We sampled vegetation and

environmental conditions throughout the landscape mosaic and asked: (i) how current land-use was

associated with vascular plant diversity and species composition; and (ii) how landscape structure

was related to total richness, richness of HNV indicator plants and richness of arable weeds.

Methods

Study area

Our study area covered approximately 7,000 km2 in the lowlands of Southern Transylvania,

Romania. The area consists of undulating terrain with altitudes from 300 to 700 m above sea level,

and its climate is subcontinental-temperate. The area comprises a mosaic of land-use types,

including arable fields (40% according to CORINE land cover), secondary grasslands and ancient

dry steppe-like grasslands (30%) and deciduous forests (30%) (Dengler et al. 2012). The natural

vegetation consists of oak-hornbeam forests (Quercus petraea-Carpinus betulus; Bohn et al. 2004).

Site selection

We followed the notion of a natural experiment (Diamond 1986), with randomised site selection in

pre-defined strata at two levels: (i) village catchments and (ii) survey sites within village catchments.

We delineated the study area into village catchments using a cost-distance algorithm that allocated

each pixel to the village with the lowest travel cost to this pixel (slope-penalized distance,

implemented in ArcGIS 10.1). We randomly selected a subset of 30 village catchments within three

different strata cross-combined by a gradient of terrain ruggedness (low, medium, high; defined by

quantiles) and protection status according to EU Birds and Habitats Directives (Site of Community

Importance (SCI), Special Protection Area (SPA) and unprotected; Table SA2.1). Within each

village catchment, we assigned land to three different land-use types using the CORINE land cover

map (EEA 2006), namely forest, grassland or arable land. By ‘arable land’, we refer to the mosaic of

arable land in its entirety, including semi-natural vegetation occurring within the mosaic, such as

Appendix II

272

field margins, road verges, hedges and old fields. Throughout grassland and arable land (collectively

termed ‘farmland’), we identified gradients of heterogeneity and woody vegetation cover. We

quantified heterogeneity as the standard deviation of panchromatic SPOT 5 data (©CNES 2007,

Distribution Spot Image SA) within a 1 ha moving window. We calculated the percentage of woody

vegetation within a 1 ha moving window by supervised classifications of the panchromatic channels

of SPOT 5, using a support vector machine algorithm (Knorn et al. 2009). We used the upper,

middle and lower thirds of these gradients to randomly select cross-replicated circular 1 ha survey

sites – 59 within arable land and 57 within grassland (Table SA2.2). An additional 23 sites (also

measuring 1 ha) were randomly selected in forest without further stratification.

Vegetation surveys

We conducted vegetation surveys between 26 May and 26 August 2012. We sampled eight plots

measuring 1 m x 1 m within each site, placed at a random distance from the center, and distributed

every 45 degrees. We alternated between random distances >40m and <40m from the center to

cover the inner and the outer 0.5 ha of the site equally. In each plot, we identified vascular plants to

species level and recorded their percent cover.

Environmental parameters

We considered variables that were potentially related to plant species richness within circles of one

hectare (henceforth: local level) and circles of 50 hectare (henceforth: landscape level) around a

given site. At the local level, we considered heterogeneity, altitude, woody vegetation cover, a heat

index (after Parker (1991): cos (slope aspect -225) * tan (slope angle)), a terrain wetness index (after

Fischer et al. 2010), and land cover type. At the landscape level, we considered terrain ruggedness,

woody vegetation cover, edge density to account for configurational heterogeneity (Fahrig et al.

2011), and a Simpson index of heterogeneity to account for compositional heterogeneity. Variables

were calculated using ArcGIS 10.1 and Fragstats 4.1, and are described in more detail in Table

A2.1.


273

Table A2.1 Definition and method description of the explanatory variables used to model plant species richness

a ©CNES (2007), ISIS programme, Distribution Spot Image SA b Based on the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Elevation Model with 30 m resolution Version 2 (GDEM V2) c CLC: Corine Land Cover Digital map 2006 d FRAGSTATS 4.2 (McGarigal et al. 2012)

Statistical analyses

The analyses consisted of three steps. First, we compared the means of alpha and beta richness

(sensu Tuomisto 2010) between the different types of land-use and different levels of protection

status. Second, we investigated patterns in community composition. Third, we modelled richness of

all plant species, richness of HNV indicator plants, and richness of arable weeds as a function of

environmental predictors. To visualize the resulting diversity patterns at a regional scale, we

predicted species richness across the entire study area.

Patterns of alpha and beta richness

For all analyses, we pooled individual plots (n = 8) within a given site (1 ha) to obtain a relative

estimate of plant species richness at the local level. First, we compared species richness between the

different land-use types (arable, grassland, forest) and different levels of protection status (SCI,

SPA and unprotected) using Analysis of Variance (ANOVA). Second, we illustrated additive beta

richness (Lande 1996) using species accumulation curves, and calculated species turnover by

additive partitioning of species richness (Veech et al. 2002; Tuomisto 2010). We tested differences

in beta richness between land-use types using ANOVA and a post-hoc Tukey test of Honest

Significant Difference (HSD).

Scale Variable Definition and method

Local (1 ha)

Heterogeneity Heterogeneity measured as the standard deviation of 2.5 m panchromatic SPOT 5 dataa

Altitude Altitude above sea level derived from ASTER DEMb Woody vegetation cover (woody 1ha)

Proportion of woody vegetation cover based on supervised classification 10m SPOT 5 dataa

Heat index Potential for ground heating calculated after Parker (1991): Heat index = cos (slope aspect -225) * tan (slope angle)

Terrain wetness index

Measure of potential soil wetness, estimated as the topographic position in the landscape and the slopeb (after Fischer et al. 2010)

Land cover Land-use classification as arable land, grassland or forest based on CLCc

Landscape (50 ha)

Ruggedness Terrain ruggedness, calculated as standard deviation of altitudeb Woody vegetation cover (woody 50ha)

Proportion of woody vegetation cover based on classified 10m SPOT 5 dataa

Edge density Configurational heterogeneity of different land covers, based on CLCc,d

Random effects

Village catchment Classification of the landscape into social-ecological units according to a cost distance algorithm of proximity to the nearest village as reference point and the slope of the terrain as cost factor

Level Observation level random effect

Appendix II

274

Species composition analysis

We conducted Detrended Correspondence Analysis (DCA) to describe species composition at the

local level and to assess its relation to possible underlying environmental gradients. We first

performed DCA including all survey sites. To more clearly capture patterns in species composition

within farmland, we performed a second DCA excluding forest sites. For both DCAs, we used a

permutation test with 1000 permutations to correlate environmental variables with the ordination.

Species richness models

We used generalized linear mixed effects models (GLMMs) to assess the effects of environmental

variables on richness of (i) all plant species, (ii) HNV indicator plants, and (iii) arable weeds. In all

cases, we specified village catchment and site (to account for overdispersion) as random effects.

Only farmland sites (i.e. arable land and grassland) were used in this analysis, because forests

differed fundamentally in structure and composition. Prior to the modelling procedure, we tested

the predictor variables for collinearity. Based on this, we excluded the Simpson index of

heterogeneity, because it was highly correlated with edge density. Furthermore, we log-transformed

local heterogeneity, local and landscape woody vegetation cover and landscape edge density to

(near-) normal distribution and standardized all continuous variables to zero mean and unit

variance. We calibrated the models using a randomly selected subset of 75% of the data. We

included interactions of land-use with all environmental variables, because we considered responses

may differ substantially between arable land and grassland. We derived the final minimum adequate

models using stepwise backward model selection (P ≤ 0.1). To validate the final models, we

predicted species richness on the remaining 25% of the dataset and compared the predicted species

richness with observed species richness using Pearson correlations.

Finally, to obtain a regional-scale overview of species richness patterns, we predicted total

plant species richness throughout the farmland of the study area, based on the fixed effect

estimates of the final model. For this purpose, we used a raster of data points with a grid size of 1

ha that lay within the calibration range of the environmental variables as measured in our field

samples (± 5%). We plotted the results in a cumulative richness density curve to visualize what

proportion of the study area’s farmland supported species richness values above or below particular

thresholds. All analyses were performed in ‘R’, using the packages vegan, ‘lme4’ and ‘bbmle’ (R

Core Team 2013).


275

Results

Diversity patterns

In total, we identified 603 vascular plant species in 139 sites, of which 25% occurred exclusively in

pastures, 20% exclusively in arable land and 7% exclusively in forests (Fig. A2.1a). Of all species,

30% occurred in more than one land-use type. We found a maximum of 50 species per 1 m × 1 m

plot (mean ± standard deviation: 14 ± 9) and a maximum of 84 species per 1 ha site (42 ± 21).

Species richness differed significantly between the three land-use types (ANOVA: F2, 136 = 81.47, P

< 0.001), but not between Natura 2000 and unprotected sites (ANOVA: F2, 136 = 0.07, P = 0.93). A

comparison of beta richness revealed significant differences between land-use types (ANOVA: F2,

136 = 81.47, P < 0.001). Species accumulation curves illustrated that grasslands had the highest beta

richness, closely followed by arable land (Fig. A2.1b).

Figure A2.1a Number of plant species that occurred exclusively in or were shared by the different landuse

types, arable land (=A), grassland (=G), and forest (=F). b: Species accumulation curves in the three different

landuse types.

Species composition

DCA ordination with all land-use types showed a clear separation of the three land-use types (Fig.

A2.2a). It revealed woody vegetation at the local and the landscape scale as the strongest underlying

environmental gradients and both correlated with the first axis (length of axis 1 = 4.92, Fig. A2.2a),

which indicated more than one full species turnover. The second axis (length = 3.48) described

gradients in topography and heterogeneity, represented by the terrain wetness index and

heterogeneity at the local scale, and ruggedness and edge density at the landscape scale. The DCA

ordination restricted to farmland sites showed a separation of grassland and arable sites, with only a

small overlap of sites (Fig. A2.2b). None of the measured environmental variables were

significantly correlated with the first axis (length = 3.51), which corresponded to approximately one

Appendix II

276

full species turnover. The second axis of the DCA (length = 2.24) correlated with landscape

variables, including both topography and configurational and compositional heterogeneity.

Figure A2.2a Detrended correspondence analysis including all survey sites; b: DCA with agricultural sites

only. Significantly correlated environmental variables are superimposed correlated environmental

superimposed (p<0.05) (Abbreviations: TWI_1ha = Local terrain wetness index; het_1ha = local

heterogeneity, ed_50_ha = landscape edge density; rugg_50ha = landscape terrain ruggedness; woody_50ha

= landscape woody vegetation cover; woody_1ha =local woody vegetation cover; SIDI_50ha = landscape

compositional heterogeneity; pd_50 ha = landscape configurational heterogeneity.

Species richness models

Independently of the land-use type, local woody vegetation cover was positively related to total

species richness (Fig. A2.3a). Total plant species richness was higher in grassland than in arable

land (Table A2.2). Terrain ruggedness at the landscape scale affected species richness positively in

arable land, but negatively in grasslands. The terrain wetness index had a negative effect on species

richness in arable land, but a slightly positive effect in grasslands (Table A2.2).

Unsurprisingly, richness of HNV indicator plants was higher in grasslands than in arable

land. Especially in arable land, richness of this group was positively related to edge density (Fig.

A2.3c). Terrain ruggedness had a positive effect on HNV plant richness in arable land and a weaker

positive effect in grassland. The terrain wetness index was negatively related to the richness of

HNV plants in arable land, but positively in grassland. We found a unimodal relationship with local

woody vegetation cover, which was negative in arable land and positive in grassland (Fig. A2.3b).

Richness of arable weeds was higher in arable land, where it decreased with increasing local woody

vegetation cover (Fig. A2.3d).


277

Figure A2.3 Results of generalized linear mixed effect models for relative richness of total species, HNV

indicator species, and arable weeds. Relationships are based on the final models summarized in Table A2.2.

Appendix II

278

Table A2.2 Results of generalized linear mixed models for total plants species richness, richness of grassland

specialists, High Nature Value (HNV) indicator plants and arable weeds. Parameter estimates are shown, with

significance levels indicated by: † P<0.1; *P<0.05; **P<0.01; ***P<0.001

Predicting total species richness across the study area highlighted that the results from the training

model correlated well with our field observation (total species richness: Pearson´s r = 0.75;

grassland specialists: r = 0.76; HNV plants: r = 0.69; arable weeds: r = 0.76). Predictions for total

plant species richness throughout the farmland mosaic of our study area resulted in predicted

values between 12 and 77 species per hectare (Fig. A2.4). The distribution of predicted values

illustrated the widespread occurrence of high species richness throughout the farmland mosaic,

with approximately half of this mosaic supporting at least 50 species per hectare (Fig. A2.5).

Response variable Total plant richness

HNV plants

Arable weeds

Intercept 3.648 0.178 0.277 Grasslands 0.453 *** 0.822 * - 2.292 *** Arable: altitude Arable: edge density 0.969 *** Arable: heat index Arable: heterogeneity Arable: ruggedness 0.215 ** 0.933 ** Arable: terrain wetness index -0.136 ** -0.347 † Arable: woody1ha -0.249 * Arable: (woody 1ha)^2 -0.722 * Arable: woody 50ha Grassland: altitude Grassland: edge density -0.918 ** Grassland: heat index Grassland: heterogeneity Grassland: ruggedness -0.234 ** -0.879 * Grassland: terrain wetness index 0.163 † 0.45 † Grassland: woody 1ha -0.106 Grassland: (woody 1ha)^2 0.931 ** Grassland: woody 50ha Altitude Edge density Heat index Ruggedness Woody 1ha 0.077 † -0.146 Woody 50ha


279

Figure A2.4 Map of predicted plant species richness for 1 ha pixels in agricultural land in the study area,

based on the final generalized linear mixed model summarized in Table A2.2. Areas that are displayed in

white were excluded from predictions, because they were either outside of the calibration range of our dataset

or represented land cover types other than farmland.

Figure A2.5 Density curves of predicted plant species richness in the farmland mosaic of the study area,

separated for arable land (black dashed line) and grassland (grey dashed line). The two lines illustrate different

distributions of species richness, which explains the two humps in the cumulative density curve (solid line).

For example, point A indicates that 50 % of the landscape contains 50 or fewer plant species per hectare.

Appendix II

280

Discussion

Besides notable exceptions (Wagner et al. 2000; Waldhardt et al. 2004; Simmering et al. 2006), this

work is one of few studies on plant diversity patterns in Europe applying a sampling approach that

covers a large extent of an agricultural landscape, and at the same time investigates environmental

variables at different spatial scales. Our results illustrate the considerable contribution of arable

land to total vascular plant diversity in a low-intensity traditional agricultural landscape. Our case

study may well reflect a situation that is typical for other Eastern European countries whose

agricultural landscapes are still structurally complex and rich in biodiversity (Young et al. 2007).

Based on our findings, we argue that maintaining the extraordinary plant diversity of low-intensity

farming landscapes calls for a conservation vision and for management plans that consider the

entire farmland mosaic.

To date, biodiversity conservation in Europe has often targeted specific sites or local

‘hotspots’ of biodiversity, instead of considering species pools across entire landscapes (Tscharntke

et al., 2012). In Romania, such hotspots are represented by extensively and traditionally managed

grasslands (Jones et al. 2010; Akeroyd & Page 2011; Dahlström et al. 2013). Some of these are

among the world’s most species-rich biotopes (Wilson et al. 2012), and often support a range of

species that are rare or endangered especially in Western European countries (Cremene et al. 2005;

Peter et al. 2009). Hence, it is not surprising that in our case study, we found 69% of all observed

plant species in grasslands. However, we also found 63% of all species were present within the

arable mosaic of our study area. Although site-level species richness was significantly lower in

arable land than in grassland, arable land also supported a considerable amount of species.

Furthermore, plant communities in arable land differed strongly from those in grasslands and

forests. In combination, these findings suggest that grassland, small-scale arable land and forests all

make important contributions to the landscape species pool. Hence, conservation management in

farmland would be most effective if it considers both arable land and grassland as integral parts of

the landscape (Matson & Vitousek 2006; Perfecto & Vandermeer 2010).

Our study unraveled several key drivers of species richness, which occasionally differed

between grassland and in arable land. In both land-use types, total species richness responded

positively to woody vegetation at the local scale. Woody vegetation may provide relatively

undisturbed refuge areas, which provide niches for several species (Ernoult & Alard 2011). Woody

vegetation therefore is an important part of structural heterogeneity, which is often lost as a result

of land-use intensification.

In grasslands, species richness was higher in locations with low ruggedness, higher terrain

wetness and higher woody vegetation cover. This might indicate the importance of continuous

areas of grasslands on species richness. Reasons why we may find lower species richness on rugged

terrain might be that (i) rugged land is vulnerable to erosion if the grazing pressure is high; and (ii)

rugged land is prone to land abandonment, because it is less accessible and less valuable for

agriculture than flat areas. Typically, abandonment induces re-growth of woody vegetation. While


281

the positive effect of a certain amount of shrub encroachment on species richness is known (and is

consistent with our findings; see above), later successional stages that are dominated by woody

vegetation do not support high grassland species diversity (Baur et al. 2006; Ruprecht et al. 2010).

Consistently with this, our findings indicate the importance of land cover heterogeneity on species

richness of HNV indicator plants. For woody vegetation and edge density, we found contrasting

effects for HNV indicator plants between arable land and grassland. Specifically, greater

heterogeneity appeared to be more beneficial in arable land than in grassland. This indicates the

importance of maintaining structural elements within the arable mosaic while simultaneously

keeping grasslands free from woody encroachment and fragmentation in order to maintain the

richness of specialized and sensitive species (see also: Zulka et al. 2014).

We presume that high plant diversity at the landscape scale in our study area results from

agricultural practices that are still dominated by semi-subsistence farming and labor-intensive,

traditional techniques, with low levels of agrochemical inputs (van Elsen 2000; Oppermann et al.

2012). In Southern Transylvania, arable land in particular contains many semi-natural elements,

including woody vegetation, which create high heterogeneity and provide different niches for a

variety of species. However, these traditional systems are being lost rapidly, and EU legislation is

likely to inadvertently foster the abandonment of traditional systems and intensification of land-use

(Kleijn & Baldi 2005; Strijker 2005; Fischer et al. 2012). It is well known from Western Europe that

modernization in the agricultural sector, including farm consolidation and agrochemical application,

has caused a drastic declines of plant species richness in arable land, with flow-on effects of other

taxa and areas (Feest et al. 2014). Application of nitrogen, for example, has been known for

decades to affect plant communities, and its negative impacts on species richness are sometimes

irreversible (Strengbom et al. 2001; Stevens et al. 2004). Given existing trends towards rural exodus,

it is likely that traditionally used grasslands in Southern Transylvania will also decrease in extent

(Government of Romania 2010). In the long term, both abandonment and intensification imply

structural simplification of the landscape, and typically cause declines in species richness in many

different taxa (Weibull et al. 2000; Benton et al. 2003; Diacon-Bolli et al. 2012; Sanderson et al.

2013).

Preventing biodiversity loss is at the core of Romania’s national rural development plan

(Government of Romania 2010). It is not only an important goal in its own right, but also

necessary to secure the delivery of important ecosystem services and maintain landscape multi-

functionality (Tscharntke et al. 2012). Moreover, protection of biodiversity in farmland is

economically far more effective than expensive restoration management in hindsight, which is

being applied in many more intensively used farmland areas in Western Europe (Kleijn et al. 2011).

To counteract loss of species diversity in agricultural landscapes, the Common Agriculture Policy

includes a series of agri-environment schemes, which provide opportunities to promote HNV

farming practices, especially in semi-natural grasslands (Page et al. 2010). Existing agri-environment

Appendix II

282

schemes, however, are largely derived from experiences in Western Europe, and their effectiveness

in new EU member states is questionable (Gorton et al. 2009; Dahlström et al. 2013).

At the moment, 20% of Romania´s farmland is of High Nature Value (European

Environment Agency 2010), and 10% is protected under Natura 2000 regulations. In our study

area, a management plan for the Natura 2000 area ‘Tarnava Mare’ is currently being developed, but

to date, it remains unclear how much effective support there will be for small-scale farmers and

other decision makers to use land in a way that promotes biodiversity. Our results indicate that

management strategies are needed that account for a range of different and contrasting effects of

environmental conditions on species richness and particular species groups. The diversity of plants

across the landscape thus appears to result from there being a wide range of different conditions

available that suit different plant species – suggesting that the mosaic character of the region as a

whole is important.

Conclusions

Despite many initiatives to conserve biodiversity in agricultural landscapes, it is highly likely that

existing policy settings, coupled with rural exodus, will cause land-use changes in many parts of

Eastern Europe. The negative effects of land-use intensification, in particular, are known to be

especially pronounced in species-rich, extensively managed agricultural landscapes (Kleijn et al.

2009), such as in Southern Transylvania. Our study showed that plants responded in a range of

different ways to variables representing structural heterogeneity, as represented by woody

vegetation cover at the local and the landscape scale, and to land-use heterogeneity at the landscape

scale. To effectively safeguard the extraordinary biodiversity of this and other biodiverse farming

landscapes of Eastern Europe, we suggest it would be useful to broaden the focus of conservation

strategies to encompass entire agricultural mosaics. Such strategies should be developed both

within and outside protected areas, and need to consider different management measures for

grassland and arable land.

Acknowledgements

The study was funded through a Sofja Kovalevskaja Award by the Alexander von Humboldt

Foundation to Joern Fischer, financed by the German Ministry for Research and Education.

Sincere thanks to the farmers, landowners, mayors and the Mihai Eminescu Trust for their support.

We warmly thank Doreen Hoffmann and Laura Sutcliffe for help in the field, Dave Abson and

Pascal Fust for analytical support and Tibor Hartel for valuable discussions.


283

References

Akeroyd, J., and S. Bădărău. 2012. Indicator plants of the High Nature Value dry grasslands of

Transylvania. Fundatia Adept, Saschiz, 16 pp.

Akeroyd, J. R., and N. Page. 2011. Conservation of High Nature Value (HNV) grassland in a

farmed landscape in Transylvania, Romania. Contributii Botanice 46:57-71.

Armengot, L., F. X. Sans, C. Fischer, A. Flohre, L. Jose-Maria, T. Tscharntke, and C. Thies. 2012.

The beta-diversity of arable weed communities on organic and conventional cereal farms

in two contrasting regions. Applied Vegetation Science 15:571-579.

Baur, B., C. Cremene, G. Groza, L. Rakosy, A. A. Schileyko, A. Baur, P. Stoll, and A. Erhardt.

2006. Effects of abandonment of subalpine hay meadows on plant and invertebrate

diversity in Transylvania, Romania. Biological Conservation 132:261-273.

Bennett, A. F., J. Q. Radford, and A. Haslem. 2006. Properties of land mosaics: implications for

nature conservation in agricultural environments. Biological Conservation 133:250-264.


the key? Trends in Ecology and Evolution 18:182-188.

Billeter, R., J. Liira, D. Bailey, R. Bugter, P. Arens, I. Augenstein, S. Aviron, J. Baudry, R. Bukacek,

F. Burel, M. Cerny, G. De Blust, R. De Cock, T. Diekötter, H. Dietz, J. Dirksen, C.

Dormann, W. Durka, M. Frenzel, R. Hamersky, F. Hendrickx, F. Herzog, S. Klotz, B.

Koolstra, A. Lausch, D. Le Coeur, J. P. Maelfait, P. Opdam, M. Roubalova, A. Schermann,

N. Schermann, T. Schmidt, O. Schweiger, M. J. M. Smulders, M. Speelmans, P. Simova, J.

Verboom, W. K. R. E. Van Wingerden, M. Zobel, and P. J. Edwards. 2008. Indicators for

biodiversity in agricultural landscapes: a pan-European study. Journal of Applied Ecology

45:141-150.

Bohn, U., G. Gollub, C. Hettwer, Z. Neuhäuslová, T. Raus, H. Schlüter, H. Weber, and S.

Hennekens,. 2004. Interactive CD-ROM map of the natural vegetation of Europe. Scale

1:2500000. Bundesamt für Naturschutz, Bonn.

Clavel, J., R. Julliard, and V. Devictor. 2010. Worldwide decline of specialist species: toward a

global functional homogenization? Frontiers in Ecology and the Environment 9:222-228.

Costanza, J. K., A. Moody, and R. K. Peet. 2011. Multi-scale environmental heterogeneity as a

predictor of plant species richness. Landscape Ecology 26:851-864.




Dahlström, A., A. Iuga, and T. Lennartsson. 2013. Managing biodiversity rich hay meadows in the

EU: a comparison of Swedish and Romanian grasslands. Environmental Conservation

40:194-205.

Davies, K. F., C. R. Margules, and J. F. Lawrence. 2004. A synergistic effect puts rare, specialized

species at greater risk of extinction. Ecology 85:265-271.

Dengler, J., T. Becker, E. Ruprecht, A. Szabó, U. Becker, M. Beldean, C. Bita-Nicolae, C. Dolnik, I.

Goia, and J. Peyrat. 2012. Festuco-Brometea communities of the Transylvanian Plateau

(Romania): a preliminary overview on syntaxonomy, ecology, and biodiversity. Tuexenia

32:319-359.

Diacon-Bolli, J., T. Daland, R. Holderegger, and M. Bürgi. 2012. Heterogeneity fosters biodiversity:

linking history and ecology of dry calcareous grasslands. Journal of Applied Ecology

13:641-653.

Diamond, J. M. 1986. Overview: laboratory experiments, field experiments, and natural

experiments. In: J. M. Diamond, and T. J. Case (eds). Community Ecology. Harper & Row,

New York, pp 3-22.

Appendix II

284

Dorresteijn, I., T. Hartel, J. Hanspach, H. von Wehrden, and J. Fischer. 2013. The conservation

value of traditional rural landscapes: the case of woodpeckers in Transylvania, Romania.

PloS ONE 8: e65236.

Ernoult, A., and D. Alard. 2011. Species richness of hedgerow habitats in changing agricultural

landscapes: are alpha and gamma diversity shaped by the same factors? Landscape Ecology

26:683-696.

European Environment Agency. 2010. Approximate distribution of HNV farmland in EU-27.

European Environment Agency, Copenhagen, Denmark.

Fahrig, L., J. Baudry, L. Brotons, F. G. Burel, T. O. Crist, R. J. Fuller, C. Sirami, G. M. Siriwardena,

and J. L. Martin. 2011. Functional landscape heterogeneity and animal biodiversity in

agricultural landscapes. Ecology Letters 14:101-112.

Feest, A., C. van Swaay, and A. van Hinsberg. 2014. Nitrogen deposition and the reduction of

butterfly biodiversity quality in the Netherlands. Ecological Indicators 39:115-119.


landscapes. Conserv Letters 5:167-175.

Fischer, J., K. Sherren, J. Stott, A. Zerger, G. Warren, and J. Stein. 2010. Toward landscape-wide

conservation outcomes in Australia's temperate grazing region. Frontiers in Ecology and

the Environment 8:69-74.

Forman, R. T. 1995. Land mosaics: the ecology of landscapes and regions. Cambridge University

Press, Cambridge, 656 pp.

Fundatia Adept. 2012. Agri-environment schemes in Romania: a success story. Fundatia Adept,

Saschiz, 21 pp.

Gabriel, D., C. Thies, and T. Tscharntke. 2005. Local diversity of arable weeds increases with

landscape complexity. Perspectives in Plant Ecology Evolution and Systematics 7:85-93.

Gorton, M., C. Hubbard, and L. Hubbard. 2009. The folly of european union policy transfer: why

the common agricultural policy (CAP) does not fit Central and Eastern Europe. Regional

Studies 43:1305-1317.

Government of Romania. 2010. National rural development programme 2007-2013. Ministry of

Agriculture and Development, Bukarest, 832 pp.

Hartel, T., and H. von Wehrden. 2013. Farmed areas predict the distribution of amphibian ponds

in a traditional rural landscape. PLoS ONE 8:e63649.

Helm, A., I. Hanski, and M. Pärtel. 2006. Slow response of plant species richness to habitat loss

and fragmentation. Ecology Letters 9:72-77.

Jones, A., J. Akeroyd, M. Beldean, and D. Turtureanu. 2010. Characterization and conservation of

xeric grasslands in the Târnava Mare area of Transylvania (Romania). Tuexenia 30:445-456.

Kleijn, D., and A. Baldi. 2005. Effects of set‐aside land on farmland biodiversity: comments on

Van Buskirk and Willi. Conservation Biology 19:963-966.





Kleijn, D., M. Rundlof, J. Scheper, H. G. Smith, and T. Tscharntke. 2011. Does conservation on


26:474-481.

Knorn, J., A. Rabe, V. C. Radeloff, T. Kuemmerle, J. Kozak, and P. Hostert. 2009. Land cover

mapping of large areas using chain classification of neighboring Landsat satellite images.

Remote Sensing of Environment 113:957-964.

Kuemmerle, T., D. Muller, P. Griffiths, and M. Rusu. 2008. Land use change in Southern Romania

after the collapse of socialism. Regional Environmental Change 9:1-12.


285

Lande, R. 1996. Statistics and partitioning of species diversity, and similarity among multiple

communities. Oikos 76:5-13.

Lindenmayer, D. B., R. B. Cunningham, C. MacGregor, M. Crane, D. Michael, J. Fischer, R.

Montague-Drake, A. Felton, and A. Manning. 2008. Temporal changes in vertebrates

during landscape transformation: a large-scale “natural experiment”. Ecological

Monographs 78:567-590.

Matson, P. A., and P. M. Vitousek. 2006. Agricultural intensification: will land spared from farming

be land spared for nature? Conservation Biology 20:709-710.

McGarigal, K., S. Cushman, and E. Ene. 2012. FRAGSTATS v4: spatial pattern analysis program

for categorical and continuous maps. Software, University of Massachusetts, Amherst.

www.umass.edu/landeco/research/fragstats/fragstats.html.

McLaughlin, A., and P. Mineau. 1995. The impact of agricultural practices on biodiversity.



and biodiversity conservation in Central Romania. Environmental Conservation 40:129-

137.

Oppermann, R., G. Beaufoy, and G. Jones 2012. High nature value farming in Europe.

Regionalkultur, Ubstadt-Weiher, 544 pp.

Page, N., R. Popa, J. Akeroyd, C. Stanciu, B. Mehedin, G. Beaufoy, K. Larkham, and L. Sutcliffe.

2010. High Nature Value grasslands: securing the ecosystem services of European farming

post 2013. Fundaţia Adept, Saschiz, 32 pp.

Parker, K. C. 1991. Topography, substrate, and vegetation patterns in the Northern Sonoran

Desert. Journal of Biogeography 18:151-163.

Perfecto, I., and J. Vandermeer. 2010. The agroecological matrix as alternative to the land-

sparing/agriculture intensification model. Proceedings of the National Academy of

Sciences 107:5786-5791.

Peter, M., A. Gigon, P. Edwards, and A. Lüscher. 2009. Changes over three decades in the floristic

composition of nutrient-poor grasslands in the Swiss Alps. Biodiversity and Conservation

18:547-567.

R Development Core Team 2013. R: a language and environment for statistical computing. R

Foundation for Statistical Computing, Vienna, Austria.

Rundlöf, M., M. Edlund, and H. G. Smith. 2010. Organic farming at local and landscape scales

benefits plant diversity. Ecography 33:514-522.

Ruprecht, E., M. Z. Enyedi, R. L. Eckstein, and T. W. Donath. 2010. Restorative removal of plant

litter and vegetation 40 years after abandonment enhances re-emergence of steppe

grassland vegetation. Biological Conservation 143:449-456.

Sanderson, F. J., M. Kucharz, M. Jobda, and P. F. Donald. 2013. Impacts of agricultural

intensification and abandonment on farmland birds in Poland following EU accession.


Simmering, D., R. Waldhardt, and A. Otte. 2006. Quantifying determinants contributing to plant

species richness in mosaic landscapes: a single- and multi-patch perspective. Landscape

Ecology 21:1233-1251.

Stevens, C. J., N. B. Dise, J. O. Mountford, and D. J. Gowing. 2004. Impact of nitrogen deposition

on the species richness of grasslands. Science 303:1876-1879.

Storkey, J., S. Meyer, K. S. Still, and C. Leuschner. 2012. The impact of agricultural intensification

and land-use change on the European arable flora. Proceedings of the Royal Society B:

Biological Sciences 279:1421-1429.

Strengbom, J., A. Nordin, T. Nasholm, and L. Ericson. 2001. Slow recovery of boreal forest

ecosystem following decreased nitrogen input. Functional Ecology 15:451-457.

Appendix II

286

Strijker, D. 2005. Marginal lands in Europe: causes of decline. Basic and Applied Ecology 6:99-106.




Tscharntke, T., J. M. Tylianakis, T. A. Rand, R. K. Didham, L. Fahrig, P. Batary, J. Bengtsson, Y.

Clough, T. O. Crist, C. F. Dormann, R. M. Ewers, J. Frund, R. D. Holt, A. Holzschuh, A.

M. Klein, D. Kleijn, C. Kremen, D. A. Landis, W. Laurance, D. Lindenmayer, C. Scherber,

N. Sodhi, I. Steffan-Dewenter, C. Thies, W. H. van der Putten, and C. Westphal. 2012.

Landscape moderation of biodiversity patterns and processes: eight hypotheses. Biological

Reviews 87:661-685.

Tuomisto, H. 2010. A consistent terminology for quantifying species diversity? Yes, it does exist.

Oecologia 164:853-860.

van Elsen, T. 2000. Species diversity as a task for organic agriculture in Europe. Agriculture,


Vandvik, V., and H. J. B. Birks. 2002. Partitioning floristic variance in Norwegian upland grasslands

into within-site and between-site components: are the patterns determined by environment

or by land-use? Plant Ecology 162:233-245.

Veech, J. A., K. S. Summerville, T. O. Crist, and J. C. Gering. 2002. The additive partitioning of

species diversity: recent revival of an old idea. Oikos 99:3-9.

Vodă, R., N. Timuș, I. Paulini, R. Popa, C. Mihali, A. Crișan, and L. Rákosy. 2010. Demographic

parameters of two sympatric Maculinea species in a Romanian site (Lepidoptera:

Lycaenidae). Entomologica Romanica 15:25-32.

Wagner, H. H., O. Wildi, and K. C. Ewald. 2000. Additive partitioning of plant species diversity in

an agricultural mosaic landscape. Landscape Ecology 15:219-227.

Waldhardt, R., D. Simmering, and A. Otte. 2004. Estimation and prediction of plant species

richness in a mosaic landscape. Landscape Ecology 19:211-226.

Weibull, A. C., J. Bengtsson, and E. Nohlgren. 2000. Diversity of butterflies in the agricultural

landscape: the role of farming system and landscape heterogeneity. Ecography 23:743-750.



Young, J., C. Richards, A. Fischer, L. Halada, T. Kull, A. Kuzniar, U. Tartes, Y. Uzunov, and A.

Watt. 2007. Conflicts between biodiversity conservation and human activities in the central

and eastern European countries. Ambio 36:545-550.

Zulka, K. P., M. Abensperg-Traun, N. Milasowszky, G. Bieringer, B.-A. Gereben-Krenn, W.

Holzinger, G. Hölzler, W. Rabitsch, A. Reischütz, P. Querner, N. Sauberer, I.

Schmitzberger, W. Willner, T. Wrbka, and H. Zechmeister. 2014. Species richness in dry

grassland patches of eastern Austria: a multi-taxon study on the role of local, landscape and

habitat quality variables. Agriculture, Ecosystems & Environment 182:25-36.


287

Supplementary Material for Appendix II

Table SA2.1 Number of villages sampled for a given combination of protection status and terrain

ruggedness

Protection statusb

No SCI SPA

TRIa Low 4 1 4

Medium 3 4 3

High 4 4 3

a Terrain ruggedness, calculated as standard deviation of altitude within one village catchment, based on the

Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) Global Digital Elevation

Model with 30 m resolution Version 2 (GDEM V2) b Protection status is according to the EU Habitats and Birds Directives

Table SA2.2 Number of survey sites (n=116) along the two local ecological gradients heterogeneity and

woody vegetation cover. A = arable land and G = grassland

Heterogeneityb low medium high

A G A G A G

Woody vegetation covera Low 8 5 7 8 0 0

Medium 6 7 8 7 7 8

High 8 7 7 8 8 7 a Proportion of woody vegetation cover based on supervised classification 10m SPOT 5 data, ©CNES

(2007), ISIS programme, Distribution Spot Image SA b Heterogeneity measured as the standard deviation of 2.5 m panchromatic SPOT 5 data, ©CNES (2007),

ISIS programme, Distribution Spot Image SA.

Appendix II

288

Table SA2.3 List of plant species used for modelling the ecological groups, namely weeds, grassland

specialists and HNV indicator species

Grassland specialists HNV indicator plants Arable weeds

Achillea millefolium Asperula cynanchica Adonis aestivalis Acinos arvensis Anthyllis vulneraria Amaranthus retroflexus Agrostis gigantea Campanula sibirica Apera spica-venti Agrostis stolonifera Coronilla varia Armoracia rusticana

Ajuga genevensis Dorycnium pentaphyllum ssp. herbaceum

Asclepias syriaca

Ajuga laxmanii Dianthus carthusianorum Avena sativa Anthyllis vulneraria Galium verum Brassica oleracea Arrhenatherum elatius Inula ensifolia Consolida regalis Artemisia campestris Linum flavum Cucurbita pepo Asperula cynanchica Linum hirsutum Echinochloa crus-galli Asperula tinctoria Onobrychis viciifolia Eruca vesicaria Astragalus monspessulanus Polygala major Fallopia convolvulus Astragalus onobrychis Scabiosa ochroleuca Galeopsis ladanum

Astrantia major Prunella grandiflora Lathyrus tuberosus Avenula pratensis Teucrium chamaedrys Oxalis corniculata Bellis perennis Thymus glabrescens Bothriochloa ischaemum Veronica austriaca Briza media Trifolium montanum

Bromus erectus Viola hirta

Bupleurum falcatum

Campanula patula

Campanula sibirica

Carex humilis

Carex pallescens

Carex spicata

Carex tomentosa

Carlina vulgaris

Centaurea jacea

Centaurea biebersteinii ssp. biebersteinii

Centaurea phrygia

Cephalaria radiata

Cerastium holosteoides

Cerastium pumilum

Chaerophyllum aromaticum

Chamaespartium sagittale

Chrysopogon gryllus

Cirsium canum

Cirsium pannonicum

Colchium autumnale

Crepis biennis

Cynosurus cristatus

Cynodon dactylon

Dactylorhiza incarnata

Danthonia alpina

Dianthus armeria

Dianthus carthusianorum

Dorycnium pentaphyllum ssp. herbaceum Elymus hispidus

Erysimum odoratum

Euphorbia cyparissias

Euphrasia officinalis

Festuca ovina

Festuca pratensis

Festuca rubra

Festuca rupicola

Festuca valesiaca

Filipendula vulgaris


289

Fragaria viridis

Galium verum

Gypsophila muralis

Helianthemum nummularium

Hieracium pilosella

Hypochoeris radicata

Juncus conglomeratus

Juncus effusus

Juncus inflexus

Juncus tenuis

Koeleria macrantha

Lathyrus nissolia

Lathyrus pratensis

Leontodon crispus ssp. crispus

Leontodon autumnalis

Leontodon hispidus

Linum austriacum

Linum catharticum

Linum flavum

Linum hirsutum

Lolium perenne

Lotus corniculatus

Luzula campestris

Lychnis flos.cuculi

Lythrum salicaria

Medicago minima

Melampyrum pratense

Nardus stricta

Oenanthe aquatica

Onobrychis viciifolia

Ononis arvensis

Orchis coriophora

Parnassia palustris

Pastinaca sativa

Phleum pratense

Pimpinella major

Plantago lanceolata

Plantago major

Poa angustifolia

Poa pratensis

Polygala comosa

Polygala major

Polygala vulgaris

Potentilla argentea

Potentilla cinerea

Potentilla erecta

Potentilla recta

Prunella laciniata

Prunella vulgaris

Ranunculus acris

Ranunculus bulbosus

Ranunculus polyanthemos

Rhinanthus rumelicus

Rhinanthus angustifolius ssp. angustifolius Rhinanthus minor

Rumex acetosa

Rumex acetosella

Salvia glutinosa

Salvia nemorosa

Salvia pratensis

Appendix II

290

Salvia transsylvanica

Sanguisorba minor

Scabiosa ochroleuca

Serratula radiata

Seseli annuum

Seseli peucedanoides

Silene nutans

Silene vulgaris

Stachys recta

Stellaria graminea

Stipa capillata

Teucrium montanum

Thalictrum lucidum

Thymus glabrescens

Thymus pannonicus

Thymus pulegioides

Tragopogon pratensis ssp. orientalis Trifolium alpestre

Trifolium arvense

Trifolium campestre

Trifolium hybridum

Trifolium montanum

Trifolium pratense

Trifolium repens

Veronica arvensis

Veronica austriaca

Veronica prostrata

Veronica spicata

Vinca herbacea

Viola canina

291

Declaration

I hereby certify that the submitted dissertation entitled ‘Biodiversity conservation in traditional

farming landscapes: The future of birds and large carnivores in Transylvania’ has been written by

me without using unauthorized aids. I did not use any aids and writings other than those indicated.

All passages taken from other writings either verbatim or in substance have been marked by me

accordingly.

I hereby confirm that in carrying out my dissertation project I have not employed the services of a

professional broker of dissertation projects, nor will I do so in the future.

This dissertation, in its present or any other version, has not yet been submitted to any other

university for review. I have not taken or registered to take another doctoral examination.

Lüneburg, 09.01.2015

_________________________________________

Ine Dorresteijn

Biodiversity conservation in traditional farming landscapesopus.uni-lueneburg.de/opus/volltexte/2015/14348/pdf/Dorresteijn... · identifying system properties that facilitate biodiversity

Documents